Psych Statistics

Psych Statistics

1. Descriptive Stats and Distributions

   • 2 DATA TYPES
     • Nominal/categorical/qualitative (eg. gender, marital status, nationality)
     • Numerical/quantitative (eg. IQ score, distance, income)
   • SCALES OF MEASUREMENT
     • Nominal (into categories)
     • Ordinal (ranked)
       • Interval (distance btwn points)
         • Ratio (same as interval but with absolute zero)
   • DATA DISTRIBUTIONS
     • Normal
     • Skewed
       • +ve (running toward x-axis)
       • -ve (running away from x-axis)
     • Kurtosis (shape of the curve)
       • Steep = most scores in the middle (leptokurtic, +ve value)
       • Flat = lots of scores at the tails (platykurtic, -ve value)
       • Normal = medium curve (mesokutic, 0 value)
     • Bi-modal (two-modes/humps)
   • MEASURES OF CENTRAL TENDENCY (the middle score)
     • Mode = most frequent score
       • ADV - reps largest no. of people, real score from data set, can be used with nominal data). DISADV - may not rep entire data set, can't be manipulated algebraically, not stable across samples as the mean.
     • Median = middle
       • ADV - not really affected by extreme scores, can be used for ordinal data. DISADV - can't be manipulated algebraically, not stable across samples as the mean
     • Mean = average
       • ADV - can be manipulated algebraically, most stable estimate across samples (better est. of population mean). DISADV - influenced by extreme scores, value might not exist in real data, can only be used with interval or ratio data
   • MEASURES OF VARIABILITY (how data varies around the middle)
     • Range = difference btwn largest and smallest value
       • Interquartile Range = diff. btwn largest and smallest value (ignoring bottom and top 25% of scores)
     • Mean Absolute Deviance - absolute values (ignore + or - of scores) of deviations then average them
       • Average Deviance - don't do this if + or - values as the scores would cancel each other out

2. 3 MAIN FUNCTIONS

   • Summarize info from multiple sources
   • Provide confidence in generalising info about something (inferential stats)
   • helps make sense of large amounts of data/info for trending (explores)

3. Population - entire set of scores research is interested in (eg. self-esteem of all people on a diet)

   • Sample - set of scores selected from the population of interest (eg. self-esteem of everyone on the paleo diet)
     • 2 SAMPLE TYPES
       • Probability - everyone has an equal chance of being sampled
         • Random Sampling - (not targeted) everyone has an equal chance of selection
         • Stratified Sampling - population divided into strata (sub groups) then random sample selected from each of the groups (eg. males and females, individuals)
         • Cluster Sampling - groups taken from population then random sample of individuals taken from clusters (eg Canb population, regional cluster, individ)
       • Non-probability - some people have no chance of being selected or likelihood can't be calculated
         • Convenience - grab (eg. asking psych students)
         • Quota - population broken into categories and participants recruited until a specific proportion is reached (eg. Psych students recruited until a number for each gender is filled)
         • Purposive - based on focus of research along
         • Snowball - based on recommendations to participate from other participants

4. ERRORS

   • Sampling Error - difference btwn sample (statistical) data and population (parameter) data
     • Standard error - Common way to measure sampling error and is the std deviation of the sampling distribution of a statistic
   • Type I - incorrectly reject null hypoth
     • Alpha: prob of type I error usually .05
   • Type II - incorrectly retain null

5. Confidence Intervals - provide an internal estimate of a parameter (range within which the parameter is likely to fall)

6. STATISTICAL SIGNIFICANCE - the alpha usually 5% (-05), very low prob of finding due to chance