Week 3: Factor analysis

Week 3: Factor analysis

1. take all the item on a q'aire and correlate them. either to one another, contructs of them or a latent construct.

   • certain items will score similarly e.g. 'I like new people' and 'I don't like being alone'
     • think about the nature of what they have in common and name it
       • overall score = too coarse. individual questions = too detailed. FACTORS = UNRELATED AS POSSIBLE, RELIABLE, HARD TO IDENTIFY

2. 4 STEPS TO FACTOR ANALYSIS: 1. decide a number of factors needed 2. construct basic factors 3. adjust so they make more sense 4. decide what the factors mean

3. communalities = variance in each item shared with other items

   • variance explained = amount of variance each component accounts for
     • scree plot = graphical representation of effectiveness of each component
       • factor loading = initial solution and rotated = how each item relates to the factor
         • naming the factors = up to the resesarcher

4. SPSS Factor analysis: one column per item.

   • OUTPUT: communalities = full item labels. indicates how much variance in each variable is explained by the analysis. extraction method stated at the bottom
   • Total variance explained: eigenvalues greater than 1 = look at. rotating sums makes slightly more similar in size. cummulative = how much it contributes
   • Scree plot = this information put into a graph. above 1 = contributes heavily.

5. factor loadings- if a variable is correlated with a factor - it belongs to that factor

   • will often appear to belong to more than 1 factor. threshhol 0.4-0.5

6. rotation = adjust factors to make more sense. tidy up the factors, easier to interpret and make sense of

7. negative loadings: variables with - loadings are still members of the factor

8. 1. carry out factor analysis with the limit on factors set to eigenvalues greater than 1. 2. if necessary, re-do FA with +/- factors based on scree plot. 3. look at items loading on each factor and give meaningful names. 4. check quality of analysis

Imágenes y archivos del mapa