Estimating replicability from language.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Can replicability be detected from the language of journal articles?\n",
    "\n",
    "In recent years, psychology has entered a replication crisis. In large part, this has been driven by the Open Science Collaboration's (2015) attempt to replicate 100 experiments published in leading journals in the field of psychology. The results are, by now, well known - only 39% of the original studies replicated, and observed effect sizes were half the magnitude of those originally observed (Open Science Collaboration, 2015).\n",
    "\n",
    "While the reasons for non replication are diverse, some key research practices are often cited as reasons for poor replication rates. These include p-hacking - the practice of continuing, or stopping, data collection until significance is reached, or slicing datasets to obtain significance from the analysis of subgroups (Head et al., 2015). Other practices include hypothesising after results are known (HARKing, Kerr, 1998), or in some instances, simply fabricating data (Simonsohn, 2013). \n",
    "\n",
    "In order to increase the robustness of scientific research, researchers have implemented a range of tools to detect errors in published research that might affect replicability. For example, the granularity-related inconsistency of means (GRIM test, Brown & Heathers, 2017) allows researchers to detect rounding errors in reported data that may or may not be related to fabricated data, or errors in reporting. Other packages like StatCheck are able to detect the plausability of reported statistics using only the text reported in published articles (Epskamp & Nuijten, 2016). \n",
    "\n",
    "Given psychology's leading role in the replication crisis, it is worth considering the role of those writing the articles - researchers themselves. Indeed, it has been suggested that part of the replication crisis is due to the pressure to publish that the academic career path places on scientists, who are driven to obtain significant results or risk unemployment (Marin-Franch, 2018). This bias towards significant results has been demonstrated through analysis of significance in published articles many times, and is well known (Head et al., 2015). The extent to which authors are actually conscious of this bias unknown.\n",
    "\n",
    "One way of examining the significance bias in science and its association with replication is to analyse the words used in the articles themselves. That is, are there patterns of language use present in articles that have been shown not to replicate, compared to those that have? If so, what words might prove to be predictors of a papers replicability? These words may indicate the language researchers invoke to 'sell' their findings, whether or not they are conscious of these biases or not. There is already evidence to suggest that the culture of academia has changed the language of scientific reporting. For example, words like 'innovative' and 'groundbreaking' have seen a relative increase of up to 15,000% between the years of 1974 and 2014 (Vinkers et al., 2015). \n",
    "\n",
    "Here, the largest body of replicated research to date is analysed - the work of the Open Science Collaboration's (2015) 100 replication attempts. Rather than examine the kinds of signifance values or reporting of descriptive statistics used elsewhere, this explorative study uses Natural Language Processing to examine the pattern of words used in both replicated and non replicated studies, and utilises machine learning algorithms to try to separate the two. If this is possible, it may be a useful approach to flag articles for likelihood of replication from their write ups alone."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Brief Method\n",
    "All data was obtained from the Open Science Framework page of the [Open Science Collaboration's replication project](https://osf.io/ezum7/). Each of the 100 journal articles listed were downloaded as PDF's from their respective journals. To work with the Python modules described below, each PDF was converted to a text file using the website [PDF to Text](https://pdftotext.com/). \n",
    "\n",
    "## Using Python to extract word corpus\n",
    "In the following notebook, a variety of Python modules are used to process the words of journal articles into a bag-of-words representation - that is, each article can be reduced to a set of words and their counts, with unnecessary words removed, such as common words such as 'The'.\n",
    "\n",
    "First, the required modules are imported, and a random seed is set for reproducibility."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Import modules\n",
    "import glob\n",
    "import nltk\n",
    "import os\n",
    "import re\n",
    "import string\n",
    "\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import seaborn as sns\n",
    "\n",
    "pd.set_option('display.max_rows', 999)\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "from collections import Counter\n",
    "from sklearn import metrics\n",
    "\n",
    "from sklearn.feature_extraction.text import CountVectorizer, TfidfTransformer\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.model_selection import train_test_split, cross_validate, GridSearchCV, LeaveOneOut, RandomizedSearchCV\n",
    "from sklearn.naive_bayes import MultinomialNB\n",
    "from sklearn.pipeline import Pipeline\n",
    "\n",
    "from scipy.stats import chi2\n",
    "from statsmodels.stats.contingency_tables import Table2x2\n",
    "from wordcloud import WordCloud\n",
    "\n",
    "# Set seed for reproducibility\n",
    "rs = np.random.seed(42)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Reading and cleaning text data\n",
    "With the data provided by the OSC, and a text file representing each paper, it is possible to programatically read in each text file and process it. In the following section and function definition, a lot of the text parsing and cleaning is carried out. The function takes the following steps on each article:\n",
    "\n",
    "* All text is converted to lower case.\n",
    "* The reference section is stripped from each article and discarded.\n",
    "* All text within parentheses are deleted - the aim of this step is to remove all in-text citations, though it is possible small sections of pertinent text are also deleted.\n",
    "* All digits are cleaned from the data - no statistical tests, descriptive statistics, counts, or page numbers are left.\n",
    "\n",
    "After the above steps, the remaining words are subjected to tokenization - splitting into sentences - as well as lemmatization. The lemmatization step reduces words to their common base form. For example, the sentence 'the experimental trials included differing stimuli' would reduce to 'experiment trial include differ stimuli'. Only sentences longer than five words were considered. \n",
    "\n",
    "There are also additional steps taken to remove common words that exist in the published articles, that are irrelevant to the reporting of the data, or other quality checks. The function definition also removes words that are less than three characters long, are the name of a journal, information about copyright, or are single letters commonly used in statistical reporting such as 'f', 'p', or 'mse'. \n",
    "\n",
    "There are also more general language processing steps taken, such as removing english stopwords and general punctuation."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Define words that should be removed from sentences, then merge to set\n",
    "journals = ['journal of experimental psychology', 'psychological science', 'journal of personality and social psychology']\n",
    "copyright_ = ['copyright', 'american psychological association', 'personal use']\n",
    "remove_words = set(journals).union(set(copyright_))\n",
    "\n",
    "# Combine english stopwords and punctuation, define words to remove, add any extras needed here\n",
    "stop_punct = set(nltk.corpus.stopwords.words('english')).union(set(string.punctuation))\n",
    "statistics_misc = ['’', '..', '...', 'f', 'p', 'mse', 'r', 't', 'et', 'al', '⫽', '⬍']\n",
    "\n",
    "# Create a word lemmatizer\n",
    "word_lem = nltk.stem.WordNetLemmatizer()\n",
    "\n",
    "# Define function to handle cleaning of text versions of PDF\n",
    "def txt_to_words(filename):\n",
    "    \n",
    "    with open(filename, 'r') as pdf:\n",
    "        \n",
    "        # Read in paper in entirety, remove all capital letters\n",
    "        paper = pdf.read().lower()\n",
    "        \n",
    "        # Remove all newline characters\n",
    "        paper = paper.replace('\\n', '')\n",
    "        \n",
    "        # Remove all text from within parentheses: This is targeting in-text citations, but may lose some actual text in parentheses, a neglible amount\n",
    "        paper = re.sub(r'\\([^()]*\\)', '', paper)\n",
    "        \n",
    "        # Remove all digits - eg statistics, page numbers\n",
    "        paper = re.sub('\\d+', '', paper)\n",
    "        \n",
    "        # Remove doi words\n",
    "        paper = re.sub(r'\\bdoi\\b', '', paper)\n",
    "        \n",
    "        # Remove the references - match the word 'references', then slice up to but not including that word\n",
    "        paper = paper[:re.search('references', paper).start()]\n",
    "            \n",
    "        # Tokenize to sentences\n",
    "        sentences = nltk.sent_tokenize(paper)\n",
    "        \n",
    "        # Check for copyright or journal name in lines - replace with empty strings if found, then continue to word tokenizing\n",
    "        word_bag = []\n",
    "        for sent in sentences:\n",
    "            \n",
    "            # Avoid the short sentences\n",
    "            if len(sent) > 5:\n",
    "                \n",
    "                # Ensure each sentence doesn't seem like the copyright or journal title name present in pdfs.\n",
    "                if not any(jargon in sent for jargon in remove_words):\n",
    "                    \n",
    "                    # Iterate over the words in this sentence\n",
    "                    for word in nltk.word_tokenize(sent):\n",
    "                        \n",
    "                        # Filter with the set and the statistics list, or very short words\n",
    "                        if word not in stop_punct and word not in statistics_misc and len(word) > 3:\n",
    "                            \n",
    "                            # If it passes, append the word, and take the final step of lemmatising it\n",
    "                            word_bag.append(word_lem.lemmatize(word))\n",
    "                           \n",
    "    return ' '.join(word_bag)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The above section details the steps to clean the text file representations of the published articles. The next section uses the raw data provided from the OSF data provided by the Open Science Collaboration to gather the filenames of articles, as well as the data about their replication status as judged by the replication team. \n",
    "\n",
    "There is some cleaning that needs implicating before any analysis can begin. The data is read into a pandas DataFrame, and several steps are taken to build in the filename of the text document. Then, the above function is applied to read in the cleaned text data from the text files, storing all information neatly in a DataFrame."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "##### Prepare DataFrame ####\n",
    "# Read in OSC data, converting Study Num column to a string\n",
    "raw = pd.read_csv('rpp_data.csv', encoding = 'latin-1', dtype = {'Study Num':str})\n",
    "\n",
    "# Subset this dataframe to extract relevant columns for analysis\n",
    "data = raw[['Study Num', 'Study Title (O)', 'Authors (O)', 'Journal (O)', 'Volume (O)', 'Issue (O)', 'Citation count, paper (O)', 'Replicate (R)']]\n",
    "\n",
    "# Of this data, some studies did not report a replication. These can be safely dropped.\n",
    "data = data[~data['Replicate (R)'].isna()]\n",
    "\n",
    "# There are duplicates in the original data, which can be identified by examining whether the study title appears more than once in the 'Study Title (O)' column\n",
    "if any(data['Study Title (O)'].duplicated()):\n",
    "    data.drop_duplicates(subset = 'Study Title (O)', inplace = True)\n",
    "    \n",
    "# The next step is to convert the 'Replicate (R)' column into a boolean digit, with a failed replication as 1\n",
    "data['Replicate (R)'] = data['Replicate (R)'].str.lower().replace({'yes':0, 'no':1})\n",
    "\n",
    "# Next, the filenames of the text files can be built from the existing columns in the DataFrame\n",
    "data['Filename'] = data['Study Num'] + '. ' + data['Study Title (O)'].str.replace('.', '') + '.txt'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Study Num</th>\n",
       "      <th>Study Title (O)</th>\n",
       "      <th>Authors (O)</th>\n",
       "      <th>Journal (O)</th>\n",
       "      <th>Volume (O)</th>\n",
       "      <th>Issue (O)</th>\n",
       "      <th>Citation count, paper (O)</th>\n",
       "      <th>Replicate (R)</th>\n",
       "      <th>Filename</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>Tracing attention and the activation flow in s...</td>\n",
       "      <td>A Roelofs</td>\n",
       "      <td>JEPLMC</td>\n",
       "      <td>34.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>46.0</td>\n",
       "      <td>1</td>\n",
       "      <td>1. Tracing attention and the activation flow i...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "      <td>Now you see it, now you don't: repetition blin...</td>\n",
       "      <td>AL Morris, ML Still</td>\n",
       "      <td>JEPLMC</td>\n",
       "      <td>34.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>1</td>\n",
       "      <td>2. Now you see it, now you don't: repetition b...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3</td>\n",
       "      <td>Working memory costs of task switching.</td>\n",
       "      <td>B Liefooghe, P Barrouillet, A Vandierendonck, ...</td>\n",
       "      <td>JEPLMC</td>\n",
       "      <td>34.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>55.0</td>\n",
       "      <td>1</td>\n",
       "      <td>3. Working memory costs of task switching.txt</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4</td>\n",
       "      <td>Accelerated relearning after retrieval-induced...</td>\n",
       "      <td>BC Storm, EL Bjork, RA Bjork</td>\n",
       "      <td>JEPLMC</td>\n",
       "      <td>34.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>39.0</td>\n",
       "      <td>1</td>\n",
       "      <td>4. Accelerated relearning after retrieval-indu...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5</td>\n",
       "      <td>The intermixed-blocked effect in human percept...</td>\n",
       "      <td>C Mitchell, S Nash, G Hall</td>\n",
       "      <td>JEPLMC</td>\n",
       "      <td>34.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>36.0</td>\n",
       "      <td>1</td>\n",
       "      <td>5. The intermixed-blocked effect in human perc...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>6</td>\n",
       "      <td>A single-system account of the relationship be...</td>\n",
       "      <td>CJ Berry, DR Shanks, RN Henson</td>\n",
       "      <td>JEPLMC</td>\n",
       "      <td>34.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>37.0</td>\n",
       "      <td>0</td>\n",
       "      <td>6. A single-system account of the relationship...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>7</td>\n",
       "      <td>Modeling distributions of immediate memory eff...</td>\n",
       "      <td>CP Beaman, I Neath, AM Surprenant</td>\n",
       "      <td>JEPLMC</td>\n",
       "      <td>34.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>12.0</td>\n",
       "      <td>1</td>\n",
       "      <td>7. Modeling distributions of immediate memory ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>8</td>\n",
       "      <td>Stereotypes and retrieval-provoked illusory so...</td>\n",
       "      <td>CS Dodson, J Darragh, A Williams</td>\n",
       "      <td>JEPLMC</td>\n",
       "      <td>34.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>14.0</td>\n",
       "      <td>1</td>\n",
       "      <td>8. Stereotypes and retrieval-provoked illusory...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>10</td>\n",
       "      <td>Across-notation automatic numerical processing.</td>\n",
       "      <td>D Ganor-Stern, J Tzelgov</td>\n",
       "      <td>JEPLMC</td>\n",
       "      <td>34.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>51.0</td>\n",
       "      <td>0</td>\n",
       "      <td>10. Across-notation automatic numerical proces...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>11</td>\n",
       "      <td>Attractor dynamics and semantic neighborhood d...</td>\n",
       "      <td>D Mirman, JS Magnuson</td>\n",
       "      <td>JEPLMC</td>\n",
       "      <td>34.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>56.0</td>\n",
       "      <td>0</td>\n",
       "      <td>11. Attractor dynamics and semantic neighborho...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>12</td>\n",
       "      <td>When does between-sequence phonological simila...</td>\n",
       "      <td>JE Marsh, F Vachon, DM Jones</td>\n",
       "      <td>JEPLMC</td>\n",
       "      <td>34.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>12.0</td>\n",
       "      <td>1</td>\n",
       "      <td>12. When does between-sequence phonological si...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>13</td>\n",
       "      <td>Bidirectional associations in multiplication m...</td>\n",
       "      <td>JI Campbell, ND Robert</td>\n",
       "      <td>JEPLMC</td>\n",
       "      <td>34.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>11.0</td>\n",
       "      <td>0</td>\n",
       "      <td>13. Bidirectional associations in multiplicati...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>15</td>\n",
       "      <td>The Stroop effect: Why proportion congruent ha...</td>\n",
       "      <td>JR Schmidt, D Besner</td>\n",
       "      <td>JEPLMC</td>\n",
       "      <td>34.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>92.0</td>\n",
       "      <td>0</td>\n",
       "      <td>15. The Stroop effect: Why proportion congruen...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>17</td>\n",
       "      <td>The ultimate sampling dilemma in experience-ba...</td>\n",
       "      <td>K Fiedler</td>\n",
       "      <td>JEPLMC</td>\n",
       "      <td>34.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>28.0</td>\n",
       "      <td>0</td>\n",
       "      <td>17. The ultimate sampling dilemma in experienc...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>19</td>\n",
       "      <td>How to say no: Single-and dual-process theorie...</td>\n",
       "      <td>K Oberauer</td>\n",
       "      <td>JEPLMC</td>\n",
       "      <td>34.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>11.0</td>\n",
       "      <td>1</td>\n",
       "      <td>19. How to say no: Single-and dual-process the...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>20</td>\n",
       "      <td>Intentional forgetting is easier after two \"sh...</td>\n",
       "      <td>L Sahakyan, PF Delaney, ER Waldum</td>\n",
       "      <td>JEPLMC</td>\n",
       "      <td>34.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>21.0</td>\n",
       "      <td>1</td>\n",
       "      <td>20. Intentional forgetting is easier after two...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>22</td>\n",
       "      <td>How does bilingualism improve executive contro...</td>\n",
       "      <td>LS Colzato, MT Bajo, W van den Wildenberg, D P...</td>\n",
       "      <td>JEPLMC</td>\n",
       "      <td>34.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>176.0</td>\n",
       "      <td>1</td>\n",
       "      <td>22. How does bilingualism improve executive co...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>24</td>\n",
       "      <td>Priming addition facts with semantic relations.</td>\n",
       "      <td>M Bassok, SF Pedigo, AT Oskarsson</td>\n",
       "      <td>JEPLMC</td>\n",
       "      <td>34.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>50.0</td>\n",
       "      <td>0</td>\n",
       "      <td>24. Priming addition facts with semantic relat...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>25</td>\n",
       "      <td>Learning correct responses and errors in the H...</td>\n",
       "      <td>M Couture, D Lafond, S Tremblay</td>\n",
       "      <td>JEPLMC</td>\n",
       "      <td>34.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>15.0</td>\n",
       "      <td>0</td>\n",
       "      <td>25. Learning correct responses and errors in t...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>26</td>\n",
       "      <td>Contextual effects on reading aloud: Evidence ...</td>\n",
       "      <td>M Reynolds, D Besner</td>\n",
       "      <td>JEPLMC</td>\n",
       "      <td>34.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>16.0</td>\n",
       "      <td>0</td>\n",
       "      <td>26. Contextual effects on reading aloud: Evide...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>27</td>\n",
       "      <td>On the additive effects of stimulus quality an...</td>\n",
       "      <td>MJ Yap, DA Balota, CS Tse, D Besner</td>\n",
       "      <td>JEPLMC</td>\n",
       "      <td>34.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>50.0</td>\n",
       "      <td>0</td>\n",
       "      <td>27. On the additive effects of stimulus qualit...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>28</td>\n",
       "      <td>Semantic interference in a delayed naming task...</td>\n",
       "      <td>N Janssen, W Schirm, BZ Mahon, A Caramazza</td>\n",
       "      <td>JEPLMC</td>\n",
       "      <td>34.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>72.0</td>\n",
       "      <td>1</td>\n",
       "      <td>28. Semantic interference in a delayed naming ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>29</td>\n",
       "      <td>Multidimensional visual statistical learning.</td>\n",
       "      <td>NB Turk-Browne, PJ Isola, BJ Scholl, TA Treat</td>\n",
       "      <td>JEPLMC</td>\n",
       "      <td>34.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>61.0</td>\n",
       "      <td>0</td>\n",
       "      <td>29. Multidimensional visual statistical learni...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>31</th>\n",
       "      <td>32</td>\n",
       "      <td>Accounting for occurrences: A new view of the ...</td>\n",
       "      <td>PA White</td>\n",
       "      <td>JEPLMC</td>\n",
       "      <td>34.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>11.0</td>\n",
       "      <td>0</td>\n",
       "      <td>32. Accounting for occurrences: A new view of ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>32</th>\n",
       "      <td>33</td>\n",
       "      <td>Multiple roles for time in short-term memory: ...</td>\n",
       "      <td>S Farrell</td>\n",
       "      <td>JEPLMC</td>\n",
       "      <td>34.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>27.0</td>\n",
       "      <td>0</td>\n",
       "      <td>33. Multiple roles for time in short-term memo...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>35</th>\n",
       "      <td>36</td>\n",
       "      <td>An attention-based associative account of adja...</td>\n",
       "      <td>S Pacton, P Perruchet</td>\n",
       "      <td>JEPLMC</td>\n",
       "      <td>34.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>72.0</td>\n",
       "      <td>0</td>\n",
       "      <td>36. An attention-based associative account of ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>36</th>\n",
       "      <td>37</td>\n",
       "      <td>Orienting attention in visual working memory r...</td>\n",
       "      <td>T Makovski, R Sussman, YV Jiang</td>\n",
       "      <td>JEPLMC</td>\n",
       "      <td>34.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>116.0</td>\n",
       "      <td>0</td>\n",
       "      <td>37. Orienting attention in visual working memo...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>38</th>\n",
       "      <td>39</td>\n",
       "      <td>Decision making and learning while taking sequ...</td>\n",
       "      <td>TJ Pleskac</td>\n",
       "      <td>JEPLMC</td>\n",
       "      <td>34.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>40.0</td>\n",
       "      <td>1</td>\n",
       "      <td>39. Decision making and learning while taking ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42</th>\n",
       "      <td>43</td>\n",
       "      <td>The rejection of moral rebels: Resenting those...</td>\n",
       "      <td>B Monin, PJ Sawyer, MJ Marquez</td>\n",
       "      <td>JPSP</td>\n",
       "      <td>95.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>88.0</td>\n",
       "      <td>1</td>\n",
       "      <td>43. The rejection of moral rebels: Resenting t...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>43</th>\n",
       "      <td>44</td>\n",
       "      <td>Why do implicit and explicit attitude tests di...</td>\n",
       "      <td>BK Payne, MA Burkley, MB Stokes</td>\n",
       "      <td>JPSP</td>\n",
       "      <td>94.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>165.0</td>\n",
       "      <td>0</td>\n",
       "      <td>44. Why do implicit and explicit attitude test...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>45</th>\n",
       "      <td>46</td>\n",
       "      <td>The developmental psychometrics of big five se...</td>\n",
       "      <td>CJ Soto, OP John, SD Gosling, J Potter</td>\n",
       "      <td>JPSP</td>\n",
       "      <td>94.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>132.0</td>\n",
       "      <td>0</td>\n",
       "      <td>46. The developmental psychometrics of big fiv...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>47</th>\n",
       "      <td>48</td>\n",
       "      <td>Terror management and adults' attachment to th...</td>\n",
       "      <td>CR Cox, J Arndt, T Pyszczynski, J Greenberg, A...</td>\n",
       "      <td>JPSP</td>\n",
       "      <td>94.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>74.0</td>\n",
       "      <td>1</td>\n",
       "      <td>48. Terror management and adults' attachment t...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>48</th>\n",
       "      <td>49</td>\n",
       "      <td>Increasing and decreasing motor and cognitive ...</td>\n",
       "      <td>D AlbarracÕn, IM Handley, K Noguchi, KC McCull...</td>\n",
       "      <td>JPSP</td>\n",
       "      <td>95.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>46.0</td>\n",
       "      <td>1</td>\n",
       "      <td>49. Increasing and decreasing motor and cognit...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>51</th>\n",
       "      <td>52</td>\n",
       "      <td>Affective incoherence: When affective concepts...</td>\n",
       "      <td>DB Centerbar, S Schnall, GL Clore, ED Garvin</td>\n",
       "      <td>JPSP</td>\n",
       "      <td>94.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>35.0</td>\n",
       "      <td>1</td>\n",
       "      <td>52. Affective incoherence: When affective conc...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>52</th>\n",
       "      <td>53</td>\n",
       "      <td>Individual differences in the regulation of in...</td>\n",
       "      <td>DM Amodio, PG Devine, E Harmon-Jones</td>\n",
       "      <td>JPSP</td>\n",
       "      <td>94.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>136.0</td>\n",
       "      <td>1</td>\n",
       "      <td>53. Individual differences in the regulation o...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>54</th>\n",
       "      <td>55</td>\n",
       "      <td>Left frontal cortical activation and spreading...</td>\n",
       "      <td>E Harmon-Jones, C Harmon-Jones, M Fearn, JD Si...</td>\n",
       "      <td>JPSP</td>\n",
       "      <td>94.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>103.0</td>\n",
       "      <td>1</td>\n",
       "      <td>55. Left frontal cortical activation and sprea...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>55</th>\n",
       "      <td>56</td>\n",
       "      <td>A social functional approach to emotions in ba...</td>\n",
       "      <td>E van Dijk, GA van Kleef, W Steinel, I van Beest</td>\n",
       "      <td>JPSP</td>\n",
       "      <td>94.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>97.0</td>\n",
       "      <td>1</td>\n",
       "      <td>56. A social functional approach to emotions i...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>57</th>\n",
       "      <td>58</td>\n",
       "      <td>Walking on eggshells: How expressing relations...</td>\n",
       "      <td>EP Lemay, MS Clark</td>\n",
       "      <td>JPSP</td>\n",
       "      <td>95.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>40.0</td>\n",
       "      <td>1</td>\n",
       "      <td>58. Walking on eggshells: How expressing relat...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>58</th>\n",
       "      <td>59</td>\n",
       "      <td>How the head liberates the heart: Projection o...</td>\n",
       "      <td>EP Lemay, MS Clark</td>\n",
       "      <td>JPSP</td>\n",
       "      <td>94.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>66.0</td>\n",
       "      <td>0</td>\n",
       "      <td>59. How the head liberates the heart: Projecti...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>60</th>\n",
       "      <td>61</td>\n",
       "      <td>Poignancy: Mixed emotional experience in the f...</td>\n",
       "      <td>H Ersner-Hershfield, JA Mikels, SJ Sullivan, L...</td>\n",
       "      <td>JPSP</td>\n",
       "      <td>94.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>80.0</td>\n",
       "      <td>1</td>\n",
       "      <td>61. Poignancy: Mixed emotional experience in t...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>62</th>\n",
       "      <td>63</td>\n",
       "      <td>1/f noise and effort on implicit measures of b...</td>\n",
       "      <td>J Correll</td>\n",
       "      <td>JPSP</td>\n",
       "      <td>94.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>42.0</td>\n",
       "      <td>1</td>\n",
       "      <td>63. 1/f noise and effort on implicit measures ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>63</th>\n",
       "      <td>64</td>\n",
       "      <td>The effect of global versus local processing s...</td>\n",
       "      <td>J F_rster, N Liberman, S Kuschel</td>\n",
       "      <td>JPSP</td>\n",
       "      <td>94.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>113.0</td>\n",
       "      <td>1</td>\n",
       "      <td>64. The effect of global versus local processi...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>64</th>\n",
       "      <td>65</td>\n",
       "      <td>Not so innocent: Does seeing one's own capacit...</td>\n",
       "      <td>JJ Exline, RF Baumeister, AL Zell, AJ Kraft, C...</td>\n",
       "      <td>JPSP</td>\n",
       "      <td>94.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>81.0</td>\n",
       "      <td>1</td>\n",
       "      <td>65. Not so innocent: Does seeing one's own cap...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>67</th>\n",
       "      <td>68</td>\n",
       "      <td>Why people are reluctant to tempt fate.</td>\n",
       "      <td>JL Risen, T Gilovich</td>\n",
       "      <td>JPSP</td>\n",
       "      <td>95.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>59.0</td>\n",
       "      <td>1</td>\n",
       "      <td>68. Why people are reluctant to tempt fate.txt</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>68</th>\n",
       "      <td>69</td>\n",
       "      <td>The nonverbal expression of pride: Evidence fo...</td>\n",
       "      <td>JL Tracy, RW Robins</td>\n",
       "      <td>JPSP</td>\n",
       "      <td>94.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>131.0</td>\n",
       "      <td>0</td>\n",
       "      <td>69. The nonverbal expression of pride: Evidenc...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>70</th>\n",
       "      <td>71</td>\n",
       "      <td>On the relative independence of thinking biase...</td>\n",
       "      <td>KE Stanovich, RF West</td>\n",
       "      <td>JPSP</td>\n",
       "      <td>94.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>291.0</td>\n",
       "      <td>1</td>\n",
       "      <td>71. On the relative independence of thinking b...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>71</th>\n",
       "      <td>72</td>\n",
       "      <td>Opening the mind to close it: Considering a me...</td>\n",
       "      <td>KL Blankenship, DT Wegener</td>\n",
       "      <td>JPSP</td>\n",
       "      <td>94.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>37.0</td>\n",
       "      <td>1</td>\n",
       "      <td>72. Opening the mind to close it: Considering ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>72</th>\n",
       "      <td>73</td>\n",
       "      <td>Distinguishing between silent and vocal minori...</td>\n",
       "      <td>KR Morrison, DT Miller</td>\n",
       "      <td>JPSP</td>\n",
       "      <td>94.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>25.0</td>\n",
       "      <td>0</td>\n",
       "      <td>73. Distinguishing between silent and vocal mi...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>76</th>\n",
       "      <td>77</td>\n",
       "      <td>Dynamics of self-regulation: How (un) accompli...</td>\n",
       "      <td>M Koo, A Fishbach</td>\n",
       "      <td>JPSP</td>\n",
       "      <td>94.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>124.0</td>\n",
       "      <td>1</td>\n",
       "      <td>77. Dynamics of self-regulation: How (un) acco...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>79</th>\n",
       "      <td>80</td>\n",
       "      <td>The effects of an implemental mind-set on atti...</td>\n",
       "      <td>MD Henderson, Y de Liver, PM Gollwitzer</td>\n",
       "      <td>JPSP</td>\n",
       "      <td>94.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>27.0</td>\n",
       "      <td>1</td>\n",
       "      <td>80. The effects of an implemental mind-set on ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>80</th>\n",
       "      <td>81</td>\n",
       "      <td>A needs-based model of reconciliation: Satisfy...</td>\n",
       "      <td>N Shnabel, A Nadler</td>\n",
       "      <td>JPSP</td>\n",
       "      <td>94.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>138.0</td>\n",
       "      <td>1</td>\n",
       "      <td>81. A needs-based model of reconciliation: Sat...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>81</th>\n",
       "      <td>82</td>\n",
       "      <td>The face of success: inferences from chief exe...</td>\n",
       "      <td>NO Rule, N Ambady</td>\n",
       "      <td>JPSP</td>\n",
       "      <td>95.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>141.0</td>\n",
       "      <td>1</td>\n",
       "      <td>82. The face of success: inferences from chief...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>83</th>\n",
       "      <td>84</td>\n",
       "      <td>Selective exposure and information quantity: H...</td>\n",
       "      <td>P Fischer, S Schulz-Hardt, D Frey</td>\n",
       "      <td>JPSP</td>\n",
       "      <td>94.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>65.0</td>\n",
       "      <td>0</td>\n",
       "      <td>84. Selective exposure and information quantit...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>85</th>\n",
       "      <td>86</td>\n",
       "      <td>Self-regulation and selective exposure: The im...</td>\n",
       "      <td>P Fischer, T Greitemeyer, D Frey</td>\n",
       "      <td>JPSP</td>\n",
       "      <td>94.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>86.0</td>\n",
       "      <td>1</td>\n",
       "      <td>86. Self-regulation and selective exposure: Th...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>86</th>\n",
       "      <td>87</td>\n",
       "      <td>The space between us: Stereotype threat and di...</td>\n",
       "      <td>PA Goff, CM Steele, PG Davies</td>\n",
       "      <td>JPSP</td>\n",
       "      <td>94.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>165.0</td>\n",
       "      <td>1</td>\n",
       "      <td>87. The space between us: Stereotype threat an...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>88</th>\n",
       "      <td>89</td>\n",
       "      <td>Sex differences in mate preferences revisited:...</td>\n",
       "      <td>PW Eastwick, EJ Finkel</td>\n",
       "      <td>JPSP</td>\n",
       "      <td>94.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>202.0</td>\n",
       "      <td>0</td>\n",
       "      <td>89. Sex differences in mate preferences revisi...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>92</th>\n",
       "      <td>93</td>\n",
       "      <td>Balancing connectedness and self-protection go...</td>\n",
       "      <td>SL Murray, JL Derrick, S Leder, JG Holmes</td>\n",
       "      <td>JPSP</td>\n",
       "      <td>94.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>117.0</td>\n",
       "      <td>1</td>\n",
       "      <td>93. Balancing connectedness and self-protectio...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>93</th>\n",
       "      <td>94</td>\n",
       "      <td>Self-handicapping, excuse making, and counterf...</td>\n",
       "      <td>SM McCrea</td>\n",
       "      <td>JPSP</td>\n",
       "      <td>95.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>51.0</td>\n",
       "      <td>0</td>\n",
       "      <td>94. Self-handicapping, excuse making, and coun...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>96</th>\n",
       "      <td>97</td>\n",
       "      <td>Social identity contingencies: How diversity c...</td>\n",
       "      <td>V Purdie-Vaughns, CM Steele, PG Davies, R Ditl...</td>\n",
       "      <td>JPSP</td>\n",
       "      <td>94.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>240.0</td>\n",
       "      <td>1</td>\n",
       "      <td>97. Social identity contingencies: How diversi...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>103</th>\n",
       "      <td>104</td>\n",
       "      <td>Effects of fluency on psychological distance a...</td>\n",
       "      <td>AL Alter, DM Oppenheimer</td>\n",
       "      <td>PS</td>\n",
       "      <td>19.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>109.0</td>\n",
       "      <td>1</td>\n",
       "      <td>104. Effects of fluency on psychological dista...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>105</th>\n",
       "      <td>106</td>\n",
       "      <td>More than meets the eye: the role of language ...</td>\n",
       "      <td>B Dessalegn, B Landau</td>\n",
       "      <td>PS</td>\n",
       "      <td>19.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>52.0</td>\n",
       "      <td>1</td>\n",
       "      <td>106. More than meets the eye: the role of lang...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>106</th>\n",
       "      <td>107</td>\n",
       "      <td>Nonconscious goal pursuit in novel environment...</td>\n",
       "      <td>B Eitam, RR Hassin, Y Schul</td>\n",
       "      <td>PS</td>\n",
       "      <td>19.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>66.0</td>\n",
       "      <td>1</td>\n",
       "      <td>107. Nonconscious goal pursuit in novel enviro...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>109</th>\n",
       "      <td>110</td>\n",
       "      <td>Perceptual mechanisms that characterize gender...</td>\n",
       "      <td>C Farris, TA Treat, RJ Viken, RM McFall</td>\n",
       "      <td>PS</td>\n",
       "      <td>19.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>37.0</td>\n",
       "      <td>1</td>\n",
       "      <td>110. Perceptual mechanisms that characterize g...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>110</th>\n",
       "      <td>111</td>\n",
       "      <td>Precision of the anchor influences the amount ...</td>\n",
       "      <td>C Janiszewski, D Uy</td>\n",
       "      <td>PS</td>\n",
       "      <td>19.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>53.0</td>\n",
       "      <td>0</td>\n",
       "      <td>111. Precision of the anchor influences the am...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>111</th>\n",
       "      <td>112</td>\n",
       "      <td>Action dynamics reveal parallel competition in...</td>\n",
       "      <td>C McKinstry, R Dale, MJ Spivey</td>\n",
       "      <td>PS</td>\n",
       "      <td>19.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>103.0</td>\n",
       "      <td>0</td>\n",
       "      <td>112. Action dynamics reveal parallel competiti...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>112</th>\n",
       "      <td>113</td>\n",
       "      <td>Prescribed optimism: Is it right to be wrong a...</td>\n",
       "      <td>DA Armor, C Massey, AM Sackett</td>\n",
       "      <td>PS</td>\n",
       "      <td>19.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>32.0</td>\n",
       "      <td>0</td>\n",
       "      <td>113. Prescribed optimism: Is it right to be wr...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>113</th>\n",
       "      <td>114</td>\n",
       "      <td>Age-related changes in the episodic simulation...</td>\n",
       "      <td>DR Addis, AT Wong, DL Schacter</td>\n",
       "      <td>PS</td>\n",
       "      <td>19.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>248.0</td>\n",
       "      <td>0</td>\n",
       "      <td>114. Age-related changes in the episodic simul...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>114</th>\n",
       "      <td>115</td>\n",
       "      <td>Preschoolers' perspective taking in word learn...</td>\n",
       "      <td>E Nurmsoo, P Bloom</td>\n",
       "      <td>PS</td>\n",
       "      <td>19.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>22.0</td>\n",
       "      <td>1</td>\n",
       "      <td>115. Preschoolers' perspective taking in word ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>115</th>\n",
       "      <td>116</td>\n",
       "      <td>Measuring the crowd within: Probabilistic repr...</td>\n",
       "      <td>E Vul, H Pashler</td>\n",
       "      <td>PS</td>\n",
       "      <td>19.0</td>\n",
       "      <td>7.0</td>\n",
       "      <td>122.0</td>\n",
       "      <td>0</td>\n",
       "      <td>116. Measuring the crowd within: Probabilistic...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>116</th>\n",
       "      <td>117</td>\n",
       "      <td>Temporal selection is suppressed, delayed, and...</td>\n",
       "      <td>E Vul, M Nieuwenstein, N Kanwisher</td>\n",
       "      <td>PS</td>\n",
       "      <td>19.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>53.0</td>\n",
       "      <td>0</td>\n",
       "      <td>117. Temporal selection is suppressed, delayed...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>117</th>\n",
       "      <td>118</td>\n",
       "      <td>Toward a physiology of dual-process reasoning ...</td>\n",
       "      <td>EJ Masicampo, RF Baumeister</td>\n",
       "      <td>PS</td>\n",
       "      <td>19.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>202.0</td>\n",
       "      <td>1</td>\n",
       "      <td>118. Toward a physiology of dual-process reaso...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>119</th>\n",
       "      <td>120</td>\n",
       "      <td>Errors Are Aversive: Defensive Motivation and ...</td>\n",
       "      <td>G Hajcak, D Foti</td>\n",
       "      <td>PS</td>\n",
       "      <td>19.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>172.0</td>\n",
       "      <td>1</td>\n",
       "      <td>120. Errors Are Aversive: Defensive Motivation...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>120</th>\n",
       "      <td>121</td>\n",
       "      <td>The sunny side of fairness: Preference for fai...</td>\n",
       "      <td>G Tabibnia, AB Satpute, MD Lieberman</td>\n",
       "      <td>PS</td>\n",
       "      <td>19.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>274.0</td>\n",
       "      <td>0</td>\n",
       "      <td>121. The sunny side of fairness: Preference fo...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>121</th>\n",
       "      <td>122</td>\n",
       "      <td>The representation of simple ensemble visual f...</td>\n",
       "      <td>GA Alvarez, A Oliva</td>\n",
       "      <td>PS</td>\n",
       "      <td>19.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>134.0</td>\n",
       "      <td>0</td>\n",
       "      <td>122. The representation of simple ensemble vis...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>123</th>\n",
       "      <td>124</td>\n",
       "      <td>Loving those who justify inequality: the effec...</td>\n",
       "      <td>GP Lau, AC Kay, SJ Spencer</td>\n",
       "      <td>PS</td>\n",
       "      <td>19.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>54.0</td>\n",
       "      <td>1</td>\n",
       "      <td>124. Loving those who justify inequality: the ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>126</th>\n",
       "      <td>127</td>\n",
       "      <td>A motion aftereffect from still photographs de...</td>\n",
       "      <td>J Winawer, AC Huk, L Boroditsky</td>\n",
       "      <td>PS</td>\n",
       "      <td>19.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>47.0</td>\n",
       "      <td>0</td>\n",
       "      <td>127. A motion aftereffect from still photograp...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>128</th>\n",
       "      <td>129</td>\n",
       "      <td>The threat of appearing prejudiced and race-ba...</td>\n",
       "      <td>JA Richeson, S Trawalter</td>\n",
       "      <td>PS</td>\n",
       "      <td>19.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>75.0</td>\n",
       "      <td>1</td>\n",
       "      <td>129. The threat of appearing prejudiced and ra...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>131</th>\n",
       "      <td>132</td>\n",
       "      <td>Where do we look during potentially offensive ...</td>\n",
       "      <td>JR Crosby, B Monin, D Richardson</td>\n",
       "      <td>PS</td>\n",
       "      <td>19.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>36.0</td>\n",
       "      <td>1</td>\n",
       "      <td>132. Where do we look during potentially offen...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>132</th>\n",
       "      <td>133</td>\n",
       "      <td>Adaptive memory: The comparative value of surv...</td>\n",
       "      <td>JS Nairne, JN Pandeirada, SR Thompson</td>\n",
       "      <td>PS</td>\n",
       "      <td>19.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>191.0</td>\n",
       "      <td>0</td>\n",
       "      <td>133. Adaptive memory: The comparative value of...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>133</th>\n",
       "      <td>134</td>\n",
       "      <td>Is happiness having what you want, wanting wha...</td>\n",
       "      <td>JT Larsen, AR McKibban</td>\n",
       "      <td>PS</td>\n",
       "      <td>19.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>22.0</td>\n",
       "      <td>0</td>\n",
       "      <td>134. Is happiness having what you want, wantin...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>134</th>\n",
       "      <td>135</td>\n",
       "      <td>Implicit attitude generalization occurs immedi...</td>\n",
       "      <td>KA Ranganath, BA Nosek</td>\n",
       "      <td>PS</td>\n",
       "      <td>19.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>65.0</td>\n",
       "      <td>1</td>\n",
       "      <td>135. Implicit attitude generalization occurs i...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>135</th>\n",
       "      <td>136</td>\n",
       "      <td>The value of believing in free will: Encouragi...</td>\n",
       "      <td>KD Vohs, JW Schooler</td>\n",
       "      <td>PS</td>\n",
       "      <td>19.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>341.0</td>\n",
       "      <td>1</td>\n",
       "      <td>136. The value of believing in free will: Enco...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>138</th>\n",
       "      <td>139</td>\n",
       "      <td>Auditory change detection: simple sounds are n...</td>\n",
       "      <td>L Demany, W Trost, M Serman, C Semal</td>\n",
       "      <td>PS</td>\n",
       "      <td>19.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>27.0</td>\n",
       "      <td>0</td>\n",
       "      <td>139. Auditory change detection: simple sounds ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>139</th>\n",
       "      <td>140</td>\n",
       "      <td>Keeping one's distance: The influence of spati...</td>\n",
       "      <td>LE Williams, JA Bargh</td>\n",
       "      <td>PS</td>\n",
       "      <td>19.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>219.0</td>\n",
       "      <td>1</td>\n",
       "      <td>140. Keeping one's distance: The influence of ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>141</th>\n",
       "      <td>142</td>\n",
       "      <td>Hedonic and instrumental motives in anger regu...</td>\n",
       "      <td>M Tamir, C Mitchell, JJ Gross</td>\n",
       "      <td>PS</td>\n",
       "      <td>19.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>138.0</td>\n",
       "      <td>0</td>\n",
       "      <td>142. Hedonic and instrumental motives in anger...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>142</th>\n",
       "      <td>143</td>\n",
       "      <td>Creating social connection through inferential...</td>\n",
       "      <td>N Epley, S Akalis, A Waytz, JT Cacioppo</td>\n",
       "      <td>PS</td>\n",
       "      <td>19.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>185.0</td>\n",
       "      <td>1</td>\n",
       "      <td>143. Creating social connection through infere...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>144</th>\n",
       "      <td>145</td>\n",
       "      <td>ñIn_Group Loveî and ñOut_Group Hateî as Motive...</td>\n",
       "      <td>N Halevy, G Bornstein, L Sagiv</td>\n",
       "      <td>PS</td>\n",
       "      <td>19.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>126.0</td>\n",
       "      <td>0</td>\n",
       "      <td>145. ñIn_Group Loveî and ñOut_Group Hateî as M...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>145</th>\n",
       "      <td>146</td>\n",
       "      <td>A word-order constraint on phonological activa...</td>\n",
       "      <td>N Janssen, FX Alario, A Caramazza</td>\n",
       "      <td>PS</td>\n",
       "      <td>19.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>15.0</td>\n",
       "      <td>1</td>\n",
       "      <td>146. A word-order constraint on phonological a...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>147</th>\n",
       "      <td>148</td>\n",
       "      <td>The Best Men Are (Not Always) Already Taken: F...</td>\n",
       "      <td>P Bressan, D Stranieri</td>\n",
       "      <td>PS</td>\n",
       "      <td>19.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>20.0</td>\n",
       "      <td>1</td>\n",
       "      <td>148. The Best Men Are (Not Always) Already Tak...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>149</th>\n",
       "      <td>150</td>\n",
       "      <td>Sensitivity to object viewpoint and action ins...</td>\n",
       "      <td>S Forti, GW Humphreys</td>\n",
       "      <td>PS</td>\n",
       "      <td>19.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>7.0</td>\n",
       "      <td>1</td>\n",
       "      <td>150. Sensitivity to object viewpoint and actio...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>150</th>\n",
       "      <td>151</td>\n",
       "      <td>With a clean conscience: Cleanliness reduces t...</td>\n",
       "      <td>S Schnall, J Benton, S Harvey</td>\n",
       "      <td>PS</td>\n",
       "      <td>19.0</td>\n",
       "      <td>12.0</td>\n",
       "      <td>246.0</td>\n",
       "      <td>1</td>\n",
       "      <td>151. With a clean conscience: Cleanliness redu...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>152</th>\n",
       "      <td>153</td>\n",
       "      <td>Extremal edge: a powerful cue to depth percept...</td>\n",
       "      <td>SE Palmer, T Ghose</td>\n",
       "      <td>PS</td>\n",
       "      <td>19.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>38.0</td>\n",
       "      <td>1</td>\n",
       "      <td>153. Extremal edge: a powerful cue to depth pe...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>153</th>\n",
       "      <td>154</td>\n",
       "      <td>What Do Cross_National Comparisons of Personal...</td>\n",
       "      <td>SJ Heine, EE Buchtel, A Norenzayan</td>\n",
       "      <td>PS</td>\n",
       "      <td>19.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>105.0</td>\n",
       "      <td>0</td>\n",
       "      <td>154. What Do Cross_National Comparisons of Per...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>154</th>\n",
       "      <td>155</td>\n",
       "      <td>Personality dominance and preferential use of ...</td>\n",
       "      <td>SK Moeller, MD Robinson, DL Zabelina</td>\n",
       "      <td>PS</td>\n",
       "      <td>19.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>33.0</td>\n",
       "      <td>1</td>\n",
       "      <td>155. Personality dominance and preferential us...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>157</th>\n",
       "      <td>158</td>\n",
       "      <td>Conflict-triggered goal shielding: Response co...</td>\n",
       "      <td>T Goschke, G Dreisbach</td>\n",
       "      <td>PS</td>\n",
       "      <td>19.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>45.0</td>\n",
       "      <td>0</td>\n",
       "      <td>158. Conflict-triggered goal shielding: Respon...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>160</th>\n",
       "      <td>161</td>\n",
       "      <td>Detecting the snake in the grass: attention to...</td>\n",
       "      <td>V Lobue, JS DeLoache</td>\n",
       "      <td>PS</td>\n",
       "      <td>19.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>143.0</td>\n",
       "      <td>1</td>\n",
       "      <td>161. Detecting the snake in the grass: attenti...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>164</th>\n",
       "      <td>165</td>\n",
       "      <td>The value heuristic in judgments of relative f...</td>\n",
       "      <td>X Dai, K Wertenbroch, CM Brendl</td>\n",
       "      <td>PS</td>\n",
       "      <td>19.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>28.0</td>\n",
       "      <td>1</td>\n",
       "      <td>165. The value heuristic in judgments of relat...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>166</th>\n",
       "      <td>167</td>\n",
       "      <td>Head up, foot down: Object words orient attent...</td>\n",
       "      <td>Z Estes, M Verges, LW Barsalou</td>\n",
       "      <td>PS</td>\n",
       "      <td>19.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>91.0</td>\n",
       "      <td>1</td>\n",
       "      <td>167. Head up, foot down: Object words orient a...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    Study Num                                    Study Title (O)  \\\n",
       "0           1  Tracing attention and the activation flow in s...   \n",
       "1           2  Now you see it, now you don't: repetition blin...   \n",
       "2           3            Working memory costs of task switching.   \n",
       "3           4  Accelerated relearning after retrieval-induced...   \n",
       "4           5  The intermixed-blocked effect in human percept...   \n",
       "5           6  A single-system account of the relationship be...   \n",
       "6           7  Modeling distributions of immediate memory eff...   \n",
       "7           8  Stereotypes and retrieval-provoked illusory so...   \n",
       "9          10    Across-notation automatic numerical processing.   \n",
       "10         11  Attractor dynamics and semantic neighborhood d...   \n",
       "11         12  When does between-sequence phonological simila...   \n",
       "12         13  Bidirectional associations in multiplication m...   \n",
       "14         15  The Stroop effect: Why proportion congruent ha...   \n",
       "16         17  The ultimate sampling dilemma in experience-ba...   \n",
       "18         19  How to say no: Single-and dual-process theorie...   \n",
       "19         20  Intentional forgetting is easier after two \"sh...   \n",
       "21         22  How does bilingualism improve executive contro...   \n",
       "23         24    Priming addition facts with semantic relations.   \n",
       "24         25  Learning correct responses and errors in the H...   \n",
       "25         26  Contextual effects on reading aloud: Evidence ...   \n",
       "26         27  On the additive effects of stimulus quality an...   \n",
       "27         28  Semantic interference in a delayed naming task...   \n",
       "28         29      Multidimensional visual statistical learning.   \n",
       "31         32  Accounting for occurrences: A new view of the ...   \n",
       "32         33  Multiple roles for time in short-term memory: ...   \n",
       "35         36  An attention-based associative account of adja...   \n",
       "36         37  Orienting attention in visual working memory r...   \n",
       "38         39  Decision making and learning while taking sequ...   \n",
       "42         43  The rejection of moral rebels: Resenting those...   \n",
       "43         44  Why do implicit and explicit attitude tests di...   \n",
       "45         46  The developmental psychometrics of big five se...   \n",
       "47         48  Terror management and adults' attachment to th...   \n",
       "48         49  Increasing and decreasing motor and cognitive ...   \n",
       "51         52  Affective incoherence: When affective concepts...   \n",
       "52         53  Individual differences in the regulation of in...   \n",
       "54         55  Left frontal cortical activation and spreading...   \n",
       "55         56  A social functional approach to emotions in ba...   \n",
       "57         58  Walking on eggshells: How expressing relations...   \n",
       "58         59  How the head liberates the heart: Projection o...   \n",
       "60         61  Poignancy: Mixed emotional experience in the f...   \n",
       "62         63  1/f noise and effort on implicit measures of b...   \n",
       "63         64  The effect of global versus local processing s...   \n",
       "64         65  Not so innocent: Does seeing one's own capacit...   \n",
       "67         68            Why people are reluctant to tempt fate.   \n",
       "68         69  The nonverbal expression of pride: Evidence fo...   \n",
       "70         71  On the relative independence of thinking biase...   \n",
       "71         72  Opening the mind to close it: Considering a me...   \n",
       "72         73  Distinguishing between silent and vocal minori...   \n",
       "76         77  Dynamics of self-regulation: How (un) accompli...   \n",
       "79         80  The effects of an implemental mind-set on atti...   \n",
       "80         81  A needs-based model of reconciliation: Satisfy...   \n",
       "81         82  The face of success: inferences from chief exe...   \n",
       "83         84  Selective exposure and information quantity: H...   \n",
       "85         86  Self-regulation and selective exposure: The im...   \n",
       "86         87  The space between us: Stereotype threat and di...   \n",
       "88         89  Sex differences in mate preferences revisited:...   \n",
       "92         93  Balancing connectedness and self-protection go...   \n",
       "93         94  Self-handicapping, excuse making, and counterf...   \n",
       "96         97  Social identity contingencies: How diversity c...   \n",
       "103       104  Effects of fluency on psychological distance a...   \n",
       "105       106  More than meets the eye: the role of language ...   \n",
       "106       107  Nonconscious goal pursuit in novel environment...   \n",
       "109       110  Perceptual mechanisms that characterize gender...   \n",
       "110       111  Precision of the anchor influences the amount ...   \n",
       "111       112  Action dynamics reveal parallel competition in...   \n",
       "112       113  Prescribed optimism: Is it right to be wrong a...   \n",
       "113       114  Age-related changes in the episodic simulation...   \n",
       "114       115  Preschoolers' perspective taking in word learn...   \n",
       "115       116  Measuring the crowd within: Probabilistic repr...   \n",
       "116       117  Temporal selection is suppressed, delayed, and...   \n",
       "117       118  Toward a physiology of dual-process reasoning ...   \n",
       "119       120  Errors Are Aversive: Defensive Motivation and ...   \n",
       "120       121  The sunny side of fairness: Preference for fai...   \n",
       "121       122  The representation of simple ensemble visual f...   \n",
       "123       124  Loving those who justify inequality: the effec...   \n",
       "126       127  A motion aftereffect from still photographs de...   \n",
       "128       129  The threat of appearing prejudiced and race-ba...   \n",
       "131       132  Where do we look during potentially offensive ...   \n",
       "132       133  Adaptive memory: The comparative value of surv...   \n",
       "133       134  Is happiness having what you want, wanting wha...   \n",
       "134       135  Implicit attitude generalization occurs immedi...   \n",
       "135       136  The value of believing in free will: Encouragi...   \n",
       "138       139  Auditory change detection: simple sounds are n...   \n",
       "139       140  Keeping one's distance: The influence of spati...   \n",
       "141       142  Hedonic and instrumental motives in anger regu...   \n",
       "142       143  Creating social connection through inferential...   \n",
       "144       145  ñIn_Group Loveî and ñOut_Group Hateî as Motive...   \n",
       "145       146  A word-order constraint on phonological activa...   \n",
       "147       148  The Best Men Are (Not Always) Already Taken: F...   \n",
       "149       150  Sensitivity to object viewpoint and action ins...   \n",
       "150       151  With a clean conscience: Cleanliness reduces t...   \n",
       "152       153  Extremal edge: a powerful cue to depth percept...   \n",
       "153       154  What Do Cross_National Comparisons of Personal...   \n",
       "154       155  Personality dominance and preferential use of ...   \n",
       "157       158  Conflict-triggered goal shielding: Response co...   \n",
       "160       161  Detecting the snake in the grass: attention to...   \n",
       "164       165  The value heuristic in judgments of relative f...   \n",
       "166       167  Head up, foot down: Object words orient attent...   \n",
       "\n",
       "                                           Authors (O) Journal (O)  \\\n",
       "0                                            A Roelofs      JEPLMC   \n",
       "1                                  AL Morris, ML Still      JEPLMC   \n",
       "2    B Liefooghe, P Barrouillet, A Vandierendonck, ...      JEPLMC   \n",
       "3                         BC Storm, EL Bjork, RA Bjork      JEPLMC   \n",
       "4                           C Mitchell, S Nash, G Hall      JEPLMC   \n",
       "5                       CJ Berry, DR Shanks, RN Henson      JEPLMC   \n",
       "6                    CP Beaman, I Neath, AM Surprenant      JEPLMC   \n",
       "7                     CS Dodson, J Darragh, A Williams      JEPLMC   \n",
       "9                             D Ganor-Stern, J Tzelgov      JEPLMC   \n",
       "10                               D Mirman, JS Magnuson      JEPLMC   \n",
       "11                        JE Marsh, F Vachon, DM Jones      JEPLMC   \n",
       "12                              JI Campbell, ND Robert      JEPLMC   \n",
       "14                                JR Schmidt, D Besner      JEPLMC   \n",
       "16                                           K Fiedler      JEPLMC   \n",
       "18                                          K Oberauer      JEPLMC   \n",
       "19                   L Sahakyan, PF Delaney, ER Waldum      JEPLMC   \n",
       "21   LS Colzato, MT Bajo, W van den Wildenberg, D P...      JEPLMC   \n",
       "23                   M Bassok, SF Pedigo, AT Oskarsson      JEPLMC   \n",
       "24                     M Couture, D Lafond, S Tremblay      JEPLMC   \n",
       "25                                M Reynolds, D Besner      JEPLMC   \n",
       "26                 MJ Yap, DA Balota, CS Tse, D Besner      JEPLMC   \n",
       "27          N Janssen, W Schirm, BZ Mahon, A Caramazza      JEPLMC   \n",
       "28       NB Turk-Browne, PJ Isola, BJ Scholl, TA Treat      JEPLMC   \n",
       "31                                            PA White      JEPLMC   \n",
       "32                                           S Farrell      JEPLMC   \n",
       "35                               S Pacton, P Perruchet      JEPLMC   \n",
       "36                     T Makovski, R Sussman, YV Jiang      JEPLMC   \n",
       "38                                          TJ Pleskac      JEPLMC   \n",
       "42                      B Monin, PJ Sawyer, MJ Marquez        JPSP   \n",
       "43                     BK Payne, MA Burkley, MB Stokes        JPSP   \n",
       "45              CJ Soto, OP John, SD Gosling, J Potter        JPSP   \n",
       "47   CR Cox, J Arndt, T Pyszczynski, J Greenberg, A...        JPSP   \n",
       "48   D AlbarracÕn, IM Handley, K Noguchi, KC McCull...        JPSP   \n",
       "51        DB Centerbar, S Schnall, GL Clore, ED Garvin        JPSP   \n",
       "52                DM Amodio, PG Devine, E Harmon-Jones        JPSP   \n",
       "54   E Harmon-Jones, C Harmon-Jones, M Fearn, JD Si...        JPSP   \n",
       "55    E van Dijk, GA van Kleef, W Steinel, I van Beest        JPSP   \n",
       "57                                  EP Lemay, MS Clark        JPSP   \n",
       "58                                  EP Lemay, MS Clark        JPSP   \n",
       "60   H Ersner-Hershfield, JA Mikels, SJ Sullivan, L...        JPSP   \n",
       "62                                           J Correll        JPSP   \n",
       "63                    J F_rster, N Liberman, S Kuschel        JPSP   \n",
       "64   JJ Exline, RF Baumeister, AL Zell, AJ Kraft, C...        JPSP   \n",
       "67                                JL Risen, T Gilovich        JPSP   \n",
       "68                                 JL Tracy, RW Robins        JPSP   \n",
       "70                               KE Stanovich, RF West        JPSP   \n",
       "71                          KL Blankenship, DT Wegener        JPSP   \n",
       "72                              KR Morrison, DT Miller        JPSP   \n",
       "76                                   M Koo, A Fishbach        JPSP   \n",
       "79             MD Henderson, Y de Liver, PM Gollwitzer        JPSP   \n",
       "80                                 N Shnabel, A Nadler        JPSP   \n",
       "81                                   NO Rule, N Ambady        JPSP   \n",
       "83                   P Fischer, S Schulz-Hardt, D Frey        JPSP   \n",
       "85                    P Fischer, T Greitemeyer, D Frey        JPSP   \n",
       "86                       PA Goff, CM Steele, PG Davies        JPSP   \n",
       "88                              PW Eastwick, EJ Finkel        JPSP   \n",
       "92           SL Murray, JL Derrick, S Leder, JG Holmes        JPSP   \n",
       "93                                           SM McCrea        JPSP   \n",
       "96   V Purdie-Vaughns, CM Steele, PG Davies, R Ditl...        JPSP   \n",
       "103                           AL Alter, DM Oppenheimer          PS   \n",
       "105                              B Dessalegn, B Landau          PS   \n",
       "106                        B Eitam, RR Hassin, Y Schul          PS   \n",
       "109            C Farris, TA Treat, RJ Viken, RM McFall          PS   \n",
       "110                                C Janiszewski, D Uy          PS   \n",
       "111                     C McKinstry, R Dale, MJ Spivey          PS   \n",
       "112                     DA Armor, C Massey, AM Sackett          PS   \n",
       "113                     DR Addis, AT Wong, DL Schacter          PS   \n",
       "114                                 E Nurmsoo, P Bloom          PS   \n",
       "115                                   E Vul, H Pashler          PS   \n",
       "116                 E Vul, M Nieuwenstein, N Kanwisher          PS   \n",
       "117                        EJ Masicampo, RF Baumeister          PS   \n",
       "119                                   G Hajcak, D Foti          PS   \n",
       "120               G Tabibnia, AB Satpute, MD Lieberman          PS   \n",
       "121                                GA Alvarez, A Oliva          PS   \n",
       "123                         GP Lau, AC Kay, SJ Spencer          PS   \n",
       "126                    J Winawer, AC Huk, L Boroditsky          PS   \n",
       "128                           JA Richeson, S Trawalter          PS   \n",
       "131                   JR Crosby, B Monin, D Richardson          PS   \n",
       "132              JS Nairne, JN Pandeirada, SR Thompson          PS   \n",
       "133                             JT Larsen, AR McKibban          PS   \n",
       "134                             KA Ranganath, BA Nosek          PS   \n",
       "135                               KD Vohs, JW Schooler          PS   \n",
       "138               L Demany, W Trost, M Serman, C Semal          PS   \n",
       "139                              LE Williams, JA Bargh          PS   \n",
       "141                      M Tamir, C Mitchell, JJ Gross          PS   \n",
       "142            N Epley, S Akalis, A Waytz, JT Cacioppo          PS   \n",
       "144                     N Halevy, G Bornstein, L Sagiv          PS   \n",
       "145                  N Janssen, FX Alario, A Caramazza          PS   \n",
       "147                             P Bressan, D Stranieri          PS   \n",
       "149                              S Forti, GW Humphreys          PS   \n",
       "150                      S Schnall, J Benton, S Harvey          PS   \n",
       "152                                 SE Palmer, T Ghose          PS   \n",
       "153                 SJ Heine, EE Buchtel, A Norenzayan          PS   \n",
       "154               SK Moeller, MD Robinson, DL Zabelina          PS   \n",
       "157                             T Goschke, G Dreisbach          PS   \n",
       "160                               V Lobue, JS DeLoache          PS   \n",
       "164                    X Dai, K Wertenbroch, CM Brendl          PS   \n",
       "166                     Z Estes, M Verges, LW Barsalou          PS   \n",
       "\n",
       "     Volume (O)  Issue (O)  Citation count, paper (O)  Replicate (R)  \\\n",
       "0          34.0        2.0                       46.0              1   \n",
       "1          34.0        1.0                        6.0              1   \n",
       "2          34.0        3.0                       55.0              1   \n",
       "3          34.0        1.0                       39.0              1   \n",
       "4          34.0        1.0                       36.0              1   \n",
       "5          34.0        1.0                       37.0              0   \n",
       "6          34.0        1.0                       12.0              1   \n",
       "7          34.0        3.0                       14.0              1   \n",
       "9          34.0        2.0                       51.0              0   \n",
       "10         34.0        1.0                       56.0              0   \n",
       "11         34.0        1.0                       12.0              1   \n",
       "12         34.0        3.0                       11.0              0   \n",
       "14         34.0        3.0                       92.0              0   \n",
       "16         34.0        1.0                       28.0              0   \n",
       "18         34.0        3.0                       11.0              1   \n",
       "19         34.0        2.0                       21.0              1   \n",
       "21         34.0        2.0                      176.0              1   \n",
       "23         34.0        2.0                       50.0              0   \n",
       "24         34.0        3.0                       15.0              0   \n",
       "25         34.0        1.0                       16.0              0   \n",
       "26         34.0        3.0                       50.0              0   \n",
       "27         34.0        1.0                       72.0              1   \n",
       "28         34.0        2.0                       61.0              0   \n",
       "31         34.0        1.0                       11.0              0   \n",
       "32         34.0        1.0                       27.0              0   \n",
       "35         34.0        1.0                       72.0              0   \n",
       "36         34.0        2.0                      116.0              0   \n",
       "38         34.0        1.0                       40.0              1   \n",
       "42         95.0        1.0                       88.0              1   \n",
       "43         94.0        1.0                      165.0              0   \n",
       "45         94.0        4.0                      132.0              0   \n",
       "47         94.0        4.0                       74.0              1   \n",
       "48         95.0        3.0                       46.0              1   \n",
       "51         94.0        4.0                       35.0              1   \n",
       "52         94.0        1.0                      136.0              1   \n",
       "54         94.0        1.0                      103.0              1   \n",
       "55         94.0        4.0                       97.0              1   \n",
       "57         95.0        2.0                       40.0              1   \n",
       "58         94.0        4.0                       66.0              0   \n",
       "60         94.0        1.0                       80.0              1   \n",
       "62         94.0        1.0                       42.0              1   \n",
       "63         94.0        4.0                      113.0              1   \n",
       "64         94.0        3.0                       81.0              1   \n",
       "67         95.0        2.0                       59.0              1   \n",
       "68         94.0        3.0                      131.0              0   \n",
       "70         94.0        4.0                      291.0              1   \n",
       "71         94.0        2.0                       37.0              1   \n",
       "72         94.0        5.0                       25.0              0   \n",
       "76         94.0        2.0                      124.0              1   \n",
       "79         94.0        3.0                       27.0              1   \n",
       "80         94.0        1.0                      138.0              1   \n",
       "81         95.0        5.0                      141.0              1   \n",
       "83         94.0        2.0                       65.0              0   \n",
       "85         94.0        3.0                       86.0              1   \n",
       "86         94.0        1.0                      165.0              1   \n",
       "88         94.0        2.0                      202.0              0   \n",
       "92         94.0        3.0                      117.0              1   \n",
       "93         95.0        2.0                       51.0              0   \n",
       "96         94.0        4.0                      240.0              1   \n",
       "103        19.0        2.0                      109.0              1   \n",
       "105        19.0        2.0                       52.0              1   \n",
       "106        19.0        3.0                       66.0              1   \n",
       "109        19.0        4.0                       37.0              1   \n",
       "110        19.0        2.0                       53.0              0   \n",
       "111        19.0        1.0                      103.0              0   \n",
       "112        19.0        4.0                       32.0              0   \n",
       "113        19.0        1.0                      248.0              0   \n",
       "114        19.0        3.0                       22.0              1   \n",
       "115        19.0        7.0                      122.0              0   \n",
       "116        19.0        1.0                       53.0              0   \n",
       "117        19.0        3.0                      202.0              1   \n",
       "119        19.0        2.0                      172.0              1   \n",
       "120        19.0        4.0                      274.0              0   \n",
       "121        19.0        4.0                      134.0              0   \n",
       "123        19.0        1.0                       54.0              1   \n",
       "126        19.0        3.0                       47.0              0   \n",
       "128        19.0        2.0                       75.0              1   \n",
       "131        19.0        3.0                       36.0              1   \n",
       "132        19.0        2.0                      191.0              0   \n",
       "133        19.0        4.0                       22.0              0   \n",
       "134        19.0        3.0                       65.0              1   \n",
       "135        19.0        1.0                      341.0              1   \n",
       "138        19.0        1.0                       27.0              0   \n",
       "139        19.0        3.0                      219.0              1   \n",
       "141        19.0        4.0                      138.0              0   \n",
       "142        19.0        2.0                      185.0              1   \n",
       "144        19.0        4.0                      126.0              0   \n",
       "145        19.0        3.0                       15.0              1   \n",
       "147        19.0        2.0                       20.0              1   \n",
       "149        19.0        1.0                        7.0              1   \n",
       "150        19.0       12.0                      246.0              1   \n",
       "152        19.0        1.0                       38.0              1   \n",
       "153        19.0        4.0                      105.0              0   \n",
       "154        19.0        4.0                       33.0              1   \n",
       "157        19.0        1.0                       45.0              0   \n",
       "160        19.0        3.0                      143.0              1   \n",
       "164        19.0        1.0                       28.0              1   \n",
       "166        19.0        2.0                       91.0              1   \n",
       "\n",
       "                                              Filename  \n",
       "0    1. Tracing attention and the activation flow i...  \n",
       "1    2. Now you see it, now you don't: repetition b...  \n",
       "2        3. Working memory costs of task switching.txt  \n",
       "3    4. Accelerated relearning after retrieval-indu...  \n",
       "4    5. The intermixed-blocked effect in human perc...  \n",
       "5    6. A single-system account of the relationship...  \n",
       "6    7. Modeling distributions of immediate memory ...  \n",
       "7    8. Stereotypes and retrieval-provoked illusory...  \n",
       "9    10. Across-notation automatic numerical proces...  \n",
       "10   11. Attractor dynamics and semantic neighborho...  \n",
       "11   12. When does between-sequence phonological si...  \n",
       "12   13. Bidirectional associations in multiplicati...  \n",
       "14   15. The Stroop effect: Why proportion congruen...  \n",
       "16   17. The ultimate sampling dilemma in experienc...  \n",
       "18   19. How to say no: Single-and dual-process the...  \n",
       "19   20. Intentional forgetting is easier after two...  \n",
       "21   22. How does bilingualism improve executive co...  \n",
       "23   24. Priming addition facts with semantic relat...  \n",
       "24   25. Learning correct responses and errors in t...  \n",
       "25   26. Contextual effects on reading aloud: Evide...  \n",
       "26   27. On the additive effects of stimulus qualit...  \n",
       "27   28. Semantic interference in a delayed naming ...  \n",
       "28   29. Multidimensional visual statistical learni...  \n",
       "31   32. Accounting for occurrences: A new view of ...  \n",
       "32   33. Multiple roles for time in short-term memo...  \n",
       "35   36. An attention-based associative account of ...  \n",
       "36   37. Orienting attention in visual working memo...  \n",
       "38   39. Decision making and learning while taking ...  \n",
       "42   43. The rejection of moral rebels: Resenting t...  \n",
       "43   44. Why do implicit and explicit attitude test...  \n",
       "45   46. The developmental psychometrics of big fiv...  \n",
       "47   48. Terror management and adults' attachment t...  \n",
       "48   49. Increasing and decreasing motor and cognit...  \n",
       "51   52. Affective incoherence: When affective conc...  \n",
       "52   53. Individual differences in the regulation o...  \n",
       "54   55. Left frontal cortical activation and sprea...  \n",
       "55   56. A social functional approach to emotions i...  \n",
       "57   58. Walking on eggshells: How expressing relat...  \n",
       "58   59. How the head liberates the heart: Projecti...  \n",
       "60   61. Poignancy: Mixed emotional experience in t...  \n",
       "62   63. 1/f noise and effort on implicit measures ...  \n",
       "63   64. The effect of global versus local processi...  \n",
       "64   65. Not so innocent: Does seeing one's own cap...  \n",
       "67      68. Why people are reluctant to tempt fate.txt  \n",
       "68   69. The nonverbal expression of pride: Evidenc...  \n",
       "70   71. On the relative independence of thinking b...  \n",
       "71   72. Opening the mind to close it: Considering ...  \n",
       "72   73. Distinguishing between silent and vocal mi...  \n",
       "76   77. Dynamics of self-regulation: How (un) acco...  \n",
       "79   80. The effects of an implemental mind-set on ...  \n",
       "80   81. A needs-based model of reconciliation: Sat...  \n",
       "81   82. The face of success: inferences from chief...  \n",
       "83   84. Selective exposure and information quantit...  \n",
       "85   86. Self-regulation and selective exposure: Th...  \n",
       "86   87. The space between us: Stereotype threat an...  \n",
       "88   89. Sex differences in mate preferences revisi...  \n",
       "92   93. Balancing connectedness and self-protectio...  \n",
       "93   94. Self-handicapping, excuse making, and coun...  \n",
       "96   97. Social identity contingencies: How diversi...  \n",
       "103  104. Effects of fluency on psychological dista...  \n",
       "105  106. More than meets the eye: the role of lang...  \n",
       "106  107. Nonconscious goal pursuit in novel enviro...  \n",
       "109  110. Perceptual mechanisms that characterize g...  \n",
       "110  111. Precision of the anchor influences the am...  \n",
       "111  112. Action dynamics reveal parallel competiti...  \n",
       "112  113. Prescribed optimism: Is it right to be wr...  \n",
       "113  114. Age-related changes in the episodic simul...  \n",
       "114  115. Preschoolers' perspective taking in word ...  \n",
       "115  116. Measuring the crowd within: Probabilistic...  \n",
       "116  117. Temporal selection is suppressed, delayed...  \n",
       "117  118. Toward a physiology of dual-process reaso...  \n",
       "119  120. Errors Are Aversive: Defensive Motivation...  \n",
       "120  121. The sunny side of fairness: Preference fo...  \n",
       "121  122. The representation of simple ensemble vis...  \n",
       "123  124. Loving those who justify inequality: the ...  \n",
       "126  127. A motion aftereffect from still photograp...  \n",
       "128  129. The threat of appearing prejudiced and ra...  \n",
       "131  132. Where do we look during potentially offen...  \n",
       "132  133. Adaptive memory: The comparative value of...  \n",
       "133  134. Is happiness having what you want, wantin...  \n",
       "134  135. Implicit attitude generalization occurs i...  \n",
       "135  136. The value of believing in free will: Enco...  \n",
       "138  139. Auditory change detection: simple sounds ...  \n",
       "139  140. Keeping one's distance: The influence of ...  \n",
       "141  142. Hedonic and instrumental motives in anger...  \n",
       "142  143. Creating social connection through infere...  \n",
       "144  145. ñIn_Group Loveî and ñOut_Group Hateî as M...  \n",
       "145  146. A word-order constraint on phonological a...  \n",
       "147  148. The Best Men Are (Not Always) Already Tak...  \n",
       "149  150. Sensitivity to object viewpoint and actio...  \n",
       "150  151. With a clean conscience: Cleanliness redu...  \n",
       "152  153. Extremal edge: a powerful cue to depth pe...  \n",
       "153  154. What Do Cross_National Comparisons of Per...  \n",
       "154  155. Personality dominance and preferential us...  \n",
       "157  158. Conflict-triggered goal shielding: Respon...  \n",
       "160  161. Detecting the snake in the grass: attenti...  \n",
       "164  165. The value heuristic in judgments of relat...  \n",
       "166  167. Head up, foot down: Object words orient a...  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Visualise the data cleaning steps so far\n",
    "display(data)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The above DataFrame appears to contain the filenames of all the text files in the current folder. The following code block ensures this is true and prevents the analysis from running if any are missing, while correcting the problematic filenames. Once this check is complete, the DataFrame is extended to include the text stored in the text files."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Generate list of text files in current directory\n",
    "text_files = glob.glob('*.txt')\n",
    "\n",
    "# Check if there are mismatches in the filename construction and text files\n",
    "if not data['Filename'].isin(text_files).all():\n",
    "    \n",
    "    # Subset the missing rows and iterate over\n",
    "    for index, entry in data[~data['Filename'].isin(text_files)].iterrows():\n",
    "        \n",
    "        # Find the match in the text file list by pairing the initial study number\n",
    "        match = [fname for fname in text_files if entry['Study Num'] + '.' in fname]\n",
    "        \n",
    "        # Replace the filename with the match\n",
    "        data.loc[index, 'Filename'] = match  \n",
    "        \n",
    "# Finally assert that Filename column matches the text file list\n",
    "assert data['Filename'].isin(text_files).all(), 'Missing entries - check text file collection!'\n",
    "\n",
    "# With above check comlete, pull in text data\n",
    "data['Text'] = data['Filename'].apply(txt_to_words)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "With the above DataFrame complete, some simple analyses can begin. \n",
    "\n",
    "## Bag of Words analysis\n",
    "A simple technique in study natural language is the bag of words approach. This converts a document into a collection of words, and a count, to represent the number of times a particular word appears. This is a simple first pass analysis that might offer insights into common words that appear more in articles that are replicated vs those that are not replicated. \n",
    "\n",
    "The next few code blocks carry out this analysis, and end with some visualisation."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Split the DataFrame into replicated and non-replicated data\n",
    "non_rep = data.loc[data['Replicate (R)'] == 1, 'Text']\n",
    "rep = data.loc[data['Replicate (R)'] == 0, 'Text']\n",
    "\n",
    "# Apply the Counter class to all articles under both replication types, splitting articles on whitespace\n",
    "non_rep_bag = Counter(word for item in non_rep for word in item.split(' '))\n",
    "rep_bag = Counter(word for item in rep for word in item.split(' '))\n",
    "\n",
    "# Compute the length of the word corporas\n",
    "len_no_rep = len(non_rep_bag)\n",
    "len_rep = len(rep_bag)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Using the `Counter` class, it is possible to extract the most common words used in the replicated and non replicated studies. The top 15 is selected here, as designated by the `n` variable.\n",
    "\n",
    "The below code block prepares these, and also creates another dictionary of shared words - words appearing in both the top `n` of each corpus. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "n = 15\n",
    "common_nonrep = {word: count for word, count in non_rep_bag.most_common(n)}\n",
    "common_rep = {word: count for word, count in rep_bag.most_common(n)}\n",
    "\n",
    "shared_words = {}\n",
    "for word, count in common_nonrep.items():\n",
    "    if word in common_rep.keys():\n",
    "        shared_words[word] = {'Failed_R': count, 'Successful_R': common_rep[word]}\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "With these dictionaries, it is possible to compute statistical differences between the word counts using a log likelihood ratio test. Using the Table2x2 class from statsmodels, it is also possible to compute p values and odds ratios. Tested against the [UCREL word frequency calculator](http://ucrel.lancs.ac.uk/llwizard.html)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "def log_likelihood(obs1, obs2, size1, size2):\n",
    "    \n",
    "    E1 = size1 * (obs1 + obs2) / (size1 + size2)\n",
    "    E2 = size2 * (obs1 + obs2) / (size1 + size2)\n",
    "\n",
    "    G2 = 2*((obs1 * np.log(obs1/E1)) + (obs2 * np.log(obs2/E2)))\n",
    "    \n",
    "    arr = np.array([[obs1, obs2], [size1, size2]])\n",
    "\n",
    "    tb = Table2x2(arr)\n",
    "\n",
    "    odds = tb.oddsratio\n",
    "    pval = tb.oddsratio_pvalue()\n",
    "\n",
    "    return G2, odds, pval"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This can now be applied to the words shared amongst the corpora in `shared_words`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Difference in frequency between failed replication and replication for\n",
      "\"participant\":\n",
      "X=378.58, OR=1.95, p=0.00000\n",
      "\n",
      "Difference in frequency between failed replication and replication for\n",
      "\"effect\":\n",
      "X=1.51, OR=1.05, p=0.23124\n",
      "\n",
      "Difference in frequency between failed replication and replication for\n",
      "\"condition\":\n",
      "X=161.11, OR=1.84, p=0.00000\n",
      "\n",
      "Difference in frequency between failed replication and replication for\n",
      "\"study\":\n",
      "X=35.72, OR=1.33, p=0.00000\n",
      "\n",
      "Difference in frequency between failed replication and replication for\n",
      "\"experiment\":\n",
      "X=14.47, OR=0.84, p=0.00018\n",
      "\n",
      "Difference in frequency between failed replication and replication for\n",
      "\"task\":\n",
      "X=14.30, OR=1.21, p=0.00022\n",
      "\n",
      "Difference in frequency between failed replication and replication for\n",
      "\"item\":\n",
      "X=0.66, OR=0.96, p=0.42306\n",
      "\n",
      "Difference in frequency between failed replication and replication for\n",
      "\"word\":\n",
      "X=8.10, OR=0.87, p=0.00487\n",
      "\n",
      "Difference in frequency between failed replication and replication for\n",
      "\"response\":\n",
      "X=12.24, OR=0.82, p=0.00052\n",
      "\n",
      "Difference in frequency between failed replication and replication for\n",
      "\"trial\":\n",
      "X=10.84, OR=0.82, p=0.00108\n",
      "\n",
      "Difference in frequency between failed replication and replication for\n",
      "\"also\":\n",
      "X=0.33, OR=0.96, p=0.57075\n",
      "\n",
      "Difference in frequency between failed replication and replication for\n",
      "\"result\":\n",
      "X=7.59, OR=0.84, p=0.00618\n",
      "\n"
     ]
    }
   ],
   "source": [
    "for word, values in shared_words.items():\n",
    "    \n",
    "    chi, odds, pval = log_likelihood(values['Failed_R'], values['Successful_R'], len_no_rep, len_rep)\n",
    "    \n",
    "    output = f'Difference in frequency between failed replication and replication for\\n\"{word}\":\\nX={chi:.2f}, OR={odds:.2f}, p={pval:.5f}\\n'\n",
    "    \n",
    "    print(output)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The above analysis reveals some interesting findings among the most commonly used words in both replicated and non-replicated research. For example, its clear that failed replications feature the words:\n",
    "* *participant*\n",
    "* *condition*\n",
    "* *study*\n",
    "* *task*\n",
    "\n",
    "more than studies that do replicate, while successfully replicated studies feature the words:\n",
    "* *result*\n",
    "* *trial*\n",
    "* *response*\n",
    "* *experiment*\n",
    "\n",
    "more than those that fail to replicate. These might indicate general language use between studies to describe conditions - e.g., do studies that replicate talk more about results and responses, while those that don't discuss numerous conditions and participant numbers? Another possibility is that these reflect the *kind* of psychology that is being written about. In the original OSC study, replication rates differed between different disciplines, with social psychology having lower rates than cognitive psychology.\n",
    "\n",
    "The below section visually highlights the above, along with some other visualisations. All numbers below represent proportions rather than absolute - values are expressed as a percentage of total copora size.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create DataFrames of the three dictionaries holding the text data\n",
    "rep_bag_df = pd.DataFrame.from_dict(common_rep, orient = 'index', columns = ['count']).reset_index()\n",
    "fail_bag_df = pd.DataFrame.from_dict(common_nonrep, orient = 'index', columns = ['count']).reset_index()\n",
    "common_df = pd.DataFrame.from_dict(shared_words, orient = 'index').reset_index()\n",
    "\n",
    "# Standardise them by the corpus size\n",
    "rep_bag_df['count'] = rep_bag_df['count'].transform(lambda x: (x/len_rep) * 100)\n",
    "fail_bag_df['count'] = fail_bag_df['count'].transform(lambda x: (x/len_no_rep) * 100)\n",
    "\n",
    "common_df['Failed_R'] = common_df['Failed_R'].transform(lambda x: (x/len_no_rep)*100)\n",
    "common_df['Successful_R'] = common_df['Successful_R'].transform(lambda x: (x/len_no_rep)*100)\n",
    "\n",
    "# Sort the common_df by failure to replicate counts\n",
    "common_df.sort_values(by='Failed_R',ascending=False,inplace=True)\n",
    "\n",
    "# Reshape common\n",
    "common = pd.melt(frame = common_df, id_vars = 'index', value_name = 'count', \n",
    "                 value_vars = ['Failed_R', 'Successful_R'], var_name = 'Replication')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1a1f7a1e48>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Lay out canvas for figures\n",
    "fig = plt.figure(figsize=(12, 12))\n",
    "grid = plt.GridSpec(2, 2, wspace=0.4, hspace=0.3)\n",
    "main = fig.add_subplot(grid[1, :2])\n",
    "left = fig.add_subplot(grid[0, 0])\n",
    "right = fig.add_subplot(grid[0, 1])\n",
    "\n",
    "# Generate comparison graph\n",
    "g = sns.barplot(data = common, y = 'index', x = 'count', hue = 'Replication', palette = ['#4d4d00', '#7a7a52'],\n",
    "            order = common.sort_values(by = 'count', ascending = False)['index'].unique(), ax = main)\n",
    "g.set(xlabel = 'Word Count (% of total)', ylabel = '', \n",
    "      title = 'Top {} most common words shared between\\nfailed and successful replications'.format(len(common['index'].unique())))\n",
    "g.set_yticklabels(g.get_yticklabels(), rotation = 30)\n",
    "g.spines['right'].set_visible(False)\n",
    "g.spines['top'].set_visible(False)\n",
    "\n",
    "# Simple count for replicated words\n",
    "g1 = sns.barplot(data = rep_bag_df, x = 'count', y = 'index', palette = ['#4d4d00'],\n",
    "                 order = rep_bag_df.sort_values(by = 'count')['index'], ax = left)\n",
    "\n",
    "g1.set(xlabel = 'Word Count (% of total)', ylabel = '',\n",
    "       title = 'Top {} words in successful replications'.format(len(rep_bag_df)))\n",
    "g1.set_yticklabels(g1.get_yticklabels(), rotation = 30)\n",
    "g1.spines['right'].set_visible(False)\n",
    "g1.spines['top'].set_visible(False)\n",
    "\n",
    "# Simple count for failed studies bag\n",
    "g2 = sns.barplot(data = fail_bag_df, x = 'count', y = 'index', palette = ['#4d4d00'],\n",
    "                 order = fail_bag_df.sort_values(by = 'count')['index'], ax = right)\n",
    "\n",
    "g2.set(xlabel = 'Word Count (% of total)', ylabel = '',\n",
    "       title = 'Top {} words in failed replications'.format(len(fail_bag_df)))\n",
    "g2.set_yticklabels(g2.get_yticklabels(), rotation = 30)\n",
    "g2.spines['right'].set_visible(False)\n",
    "g2.spines['top'].set_visible(False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Machine learning approaches\n",
    "The above simple analysis made clear that papers that failed to replicate and those that successfully replicated, while sharing broadly similar common words, nonetheless have significant differences in terms of the kinds of words used. This warrants a further exploration of the dataset.\n",
    "\n",
    "There are a number of machine learning (ML) algorithms that are designed to classify texts based on certain criteria. This is achieved by decomposing documents into vectors comprising the counts of words, which can be raw counts or term-frequency inverse document frequency weightings - i.e., where words are weighted by their relatively scarcity in a body of documents. These vectors can act as predictors to a dependent variable in a classifier algorithm. Some of these will be implemented below."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "# First, split apart the DataFrame into the predictors and dependent variables\n",
    "y = data['Replicate (R)']\n",
    "X = data['Text']\n",
    "\n",
    "# Conduct a train-test split to create a hold out set for prediction later on - keep 10% of data back, given the size of data\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = 0.10,random_state=rs)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Naive Bayes Classifier\n",
    "The first ML algorithm to try to classify the dataset is a Naive Bayes classifier. NB works by....\n",
    "\n",
    "The vectorizer class also allows for searching for n_grams - searching for words that appear next to one another commonly. These might also reveal what are important predictors of documents being classified. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Set up processing pipeline for NB\n",
    "# Vectorize, then allow for a TF-IDF transform, before applying the classifier over the data\n",
    "nb_pipe = Pipeline([('vectorizer', CountVectorizer(min_df = 2, stop_words = 'english')),\n",
    "                    ('transformer', TfidfTransformer()),\n",
    "                    ('classifier', MultinomialNB()) ])\n",
    "\n",
    "# Prepare the pipeline parameters, looking for n_grams\n",
    "naive_b_params = {'vectorizer__ngram_range':[(1, 2), (1, 3), (1, 4), (1, 5)],\n",
    "            'transformer__use_idf':[True, False],\n",
    "            'classifier__alpha':[0.001, 0.01, 0.1, 1]}"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Using the above processing pipeline, it is possible to conduct a grid search, trying various combinations of the parameters (e.g., varying n gram lengths, using raw counts instead of TF-IDF, and the amount of smoothing the model uses) to find the optimal predictive value, before trying to predict the hold out data. This is implemented below."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 5 folds for each of 32 candidates, totalling 160 fits\n"
     ]
    }
   ],
   "source": [
    "# Make a grid search object\n",
    "nb_grid = GridSearchCV(nb_pipe,naive_b_params,scoring='accuracy',cv=5,n_jobs=-1,verbose=1)\n",
    "\n",
    "# Fit to training data - this will take some time\n",
    "nb_grid.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Examine some of the findings!\n",
    "# What was the best classifier?\n",
    "print('Best params: ', nb_grid.best_params_)\n",
    "\n",
    "# How well did the classifier perform on the training data?\n",
    "print('Training data best accuracy: {:.2f}'.format(nb_grid.best_score_))\n",
    "\n",
    "# What proportion of failed replications are in the training data - e.g., if classifier just guessed 'one' all the time, what would it give us?\n",
    "print('Proportion of failed replications in training: {:.2f}'.format(y_train.mean()))\n",
    "\n",
    "# How does it do on the test set?\n",
    "print('Test set accuracy: {:.2f}'.format(nb_grid.score(X_test, y_test)))\n",
    "\n",
    "# Finally, what proportion of failed reps are in the test data?\n",
    "print('Proportion of failed replications in test: {:.2f}'.format(y_test.mean()))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "80% accuracy is fairly good for a task like this. It can be compared with another classifier, like a logistic regression. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Logistic Regression\n",
    "This can be set up very similarly to the Naive Bayes."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Set up processing pipeline for NB\n",
    "# Vectorize, then allow for a TF-IDF transform, before applying the classifier over the data\n",
    "logreg_pipe = Pipeline([('vectorizer', CountVectorizer(min_df = 2, stop_words = 'english')),\n",
    "                    ('transformer', TfidfTransformer()),\n",
    "                    ('classifier', LogisticRegression()) ])\n",
    "\n",
    "# Prepare the pipeline parameters, looking for n_grams\n",
    "logreg_params = {'vectorizer__ngram_range':[(1, 2), (1, 3), (1, 4), (1, 5)],\n",
    "            'transformer__use_idf':[True, False],\n",
    "            'classifier__C':[0.001, 0.01, 0.1, 1]}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Make a grid search object\n",
    "logreg_grid = GridSearchCV(logreg_pipe,logreg_params,scoring='accuracy',cv=5,n_jobs=-1,verbose=1)\n",
    "\n",
    "# Fit to training data - this will take some time\n",
    "logreg_grid.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Examine some of the findings!\n",
    "# What was the best classifier?\n",
    "print('Best params: ', logreg_grid.best_params_)\n",
    "\n",
    "# How well did the classifier perform on the training data?\n",
    "print('Training data best accuracy: {:.2f}'.format(logreg_grid.best_score_))\n",
    "\n",
    "# What proportion of failed replications are in the training data - e.g., if classifier just guessed 'one' all the time, what would it give us?\n",
    "print('Proportion of failed replications in training: {:.2f}'.format(y_train.mean()))\n",
    "\n",
    "# How does it do on the test set?\n",
    "print('Test set accuracy: {:.2f}'.format(logreg_grid.score(X_test, y_test)))\n",
    "\n",
    "# Finally, what proportion of failed reps are in the test data?\n",
    "print('Proportion of failed replications in test: {:.2f}'.format(y_test.mean()))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The logistic regression doesn't do as well as the Naive Bayes. We can explore the Naive Bayes in more detail, examining feature importances."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Digging into Naive Bayes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# What data is the naive bayes using to make a decision about failed replication?\n",
    "# First, retrieve the best estimator from the grid search object\n",
    "NB_best = nb_grid.best_estimator_\n",
    "\n",
    "# Retrieve the word list used by the CountVectorizer\n",
    "words = NB_best.named_steps['vectorizer'].get_feature_names()\n",
    "\n",
    "# Extract the probabilities per word for rep and non rep, sorted by importance\n",
    "success_probs = NB_best.named_steps['classifier'].feature_log_prob_[0,:].argsort()\n",
    "failure_probs = NB_best.named_steps['classifier'].feature_log_prob_[1,:].argsort()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Print words associated with the highest and lowest probabilities\n",
    "replicate = np.take(words, success_probs[-20:]).tolist()\n",
    "failure = np.take(words, failure_probs[-20:]).tolist()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Show word clouds\n",
    "rep = WordCloud().generate(' '.join(replicate))\n",
    "fail = WordCloud().generate(' '.join(failure))\n",
    "\n",
    "f, (ax1, ax2) = plt.subplots(1,2,figsize=(12,12))\n",
    "ax1.imshow(rep)\n",
    "ax1.set_title('Highly weighted words for successful replication')\n",
    "ax1.axis('off')\n",
    "\n",
    "ax2.imshow(fail)\n",
    "ax2.set_title('Highly weighted words for failed replication')\n",
    "ax2.axis('off')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Examine what the model does right and wrong through confusion matrix\n",
    "y_pred = nb_grid.predict(X_test)\n",
    "nb_conf_mat = metrics.classification_report(y_test, y_pred, target_names=['Successful', 'Failure'])\n",
    "print(metrics.roc_auc_score(y_test, y_pred))\n",
    "\n",
    "y_pred = logreg_grid.predict(X_test)\n",
    "logreg_conf_mat = metrics.classification_report(y_test, y_pred, target_names=['Successful', 'Failure'])\n",
    "\n",
    "print('Naive Bayes Classifier\\n',nb_conf_mat)\n",
    "print('Logistic Regression Classifier\\n',logreg_conf_mat)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "It's clear the model does a good job of predicting the failed replications, but not so much the successful ones. It's therefore important to conduct a nested cross validation, which will reveal the potential ability of the model to classify across a range of splits - it might just be the initial train/test split causing high recall."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "scoring = ['precision', 'recall']\n",
    "\n",
    "nb_grid2 = GridSearchCV(nb_pipe,naive_b_params,cv=10)\n",
    "\n",
    "scores = cross_validate(NB_best, X, y, scoring=['recall','precision', 'f1'],cv=10,n_jobs=-1,return_train_score=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Print mean scores\n",
    "print('Mean cross validated precision: {}'.format(scores['test_precision'].mean()))\n",
    "print('Mean cross validated recall: {}'.format(scores['test_recall'].mean()))\n",
    "print('Mean cross validated F1 score: {}'.format(scores['test_f1'].mean()))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To conclude, the classifier does only moderately well in predicting replicability from language. While recall is high - i.e., the classifier is able to quite accurately detect the failed replications, it does not do so well on detecting the successful replications. In some ways this is not so much of a problem, as detecting studies likely to fail a replication attempt should be found over those that do not. \n",
    "\n",
    "However, the words that the model seems to use to discriminate are often linked to the kind of psychology being done, such as social versus cognitive. This supports the findings of the OSC (2015) project, where cognitive psychology had relatively higher rates of replication compared to social psychology.\n",
    "\n",
    "#### Conclusion\n",
    "This mini-project has utilised the raw currency of academic literature: published papers themselves. Using natural language processing tools and relatively simple machine learning approaches, a classifier could successfully recall experiments that had failed to replicate from the language they used alone, corroborating the findings of the OSC (2015) project. With more training data after more replication attempts, accuracy may increase and be able to discriminate with greater accuracy without relying on the kind of psychology used. "
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}