What does component signifies in Principal component analysis?

[rahul_panthri]

Structure Matrix

......................................Component

...................................1 ..... 2 .....3 ......4

Eff_class.................. .893 .139 .236 .143

Prod_features.......... .889 -.194 .209 .219

offer......................... .912 .006 .486 .278

Brand_recognition... .183 .057 .905 .243

Price....................... -.015 .939 .106 .146

Service_network..... ...643 .163 .766 .633

Pre_sales_support.. .839 .124 .811 .275

After_sales_support.. .506 .301 .478 .760

On_times_delivery... -.079 .938 -.018 -.240

Consultant_recn...... .026 -.164 .147 .897

Extraction Method: Principal Component Analysis.

Rotation Method: Promax with Kaiser Normalization.

I have done the Principal component analysis. In my survey, I asked the question "How much weight-age would you give to these factors before making purchase decision on a level 1 to 5 with 5 being highest" to determine the factors affecting purchase decision. Now what i make out of a analysis is that despite the presumption that purchase decision is the latent construct there are 4 different latent variables in between or the components on which group of factors are loading which could be seen through pattern matrix but how should i name these components. What should i write in my analysis for the report. For Ex. Efficiency class, product features and Discount offers(First three factors in table) seems to be correlated with 1st component though in reality they can't be grouped into anything which could be assigned a name. Thanks in advance

 

[Miner]

You seem to be going across the rows. You should go down the columns. Look at the attached heat map. While not really a statistical tool, it does help you visualize. The first component is mainly comprised of Eff. Class, Prod_Features, offer and Pre_Sales_Support. This is what you try to give a name. The second component is Price and On_time_Delivery. Name that.

 

[noetsi]

It is normal in EFA to rotate your factors before you name it (and I thought this was true of PCA as well). That makes naming anything easier commonly.

 

[rahul_panthri]

Thanks for the reply. but how can product features, Discount offers and Sales support be named together as they are different facets altogether.

 

[rahul_panthri]

You seem to be going across the rows. You should go down the columns. Look at the attached heat map. While not really a statistical tool, it does help you visualize. The first component is mainly comprised of Eff. Class, Prod_Features, offer and Pre_Sales_Support. This is what you try to give a name. The second component is Price and On_time_Delivery. Name that.

Thanks for the reply, but how can Product features, Discount offers and Sales support be named together.

 

[Miner]

Typically, when you perform this type of analysis, you have some expert domain knowledge upon which to rely. I do not have that, but I'll take a stab at it.

Product features, discount offers and sales support might be named as "Pre-Sale Value". Service network, after sales support and consultant_rcn? might be "Post sales value". Again, I am speculating, because I don't know what all of the terms mean.