Chris
Snijders
© 2012 Chris Snijders Contact Me

Social orientation measurement using 1 question


The Social Orientation Value (SVO) is a measurement of someone’s “social preferences”. Literally, it is meant to measure the extent to which you care about what others get. Often, It is measured through the “RING measure” (a tedious procedure with many questions), or through a series of decomposed games. In this latter method, you typically get 9 comparisons and are classified as egoist, altruist or competitor if you choose consistent with that label at least 6 out of 9 times.

I often come across papers or presentations where SVO is measured, and usually I manage to remember that a long while ago, Jeroen Weesie and me came up with a single question measure for the social orientation value (as part of the HIN95Exp booklet). We published about this in a Dutch book, Kalmijn et al. (1999) Huwelijks- en samenwoonrelaties in Nederland. De organisatie van afhankelijkheid. ISBN13: 9789023234524

If I remember correctly, we had a chapter in there about the effects of having (dis)similar time preferences and (dis)similar social value orientations on the extent of marital problems.

Unfortunately, although I remembered how we did it, I could not find the data back. Today I did! It worked as follows. The single question we used was:


We present you a situation in which you have to choose between 4 possible payments to you and a random other participant in this survey:

you get--------the other gets--------your choice
1)  69 24  .
2)  61 55  .
3)  78 13 .
4)  64 34 .

We like to know how attractive you consider each of these possibilities. Look careful at the possibilities. Put a "1" behind the option that you consider most attractive. Now, put the number "4" behind the option that you find the least attractive. Now put a "2" behind the option that you find most attractive of the two remaining options. Finally, put a "3" behind the one option that is left.

The smart bit about it is that it is actually asking for six comparisons simultaneously. Assuming that the SVO-value equals x, you prefer the first over the second option if 69+24x > 61+55x, which happens if and only if x<8/31. The six implicit comparisons are (I hope I did the math right):

1>2 <=> x<8/31 => threshold at 0.26
1>3 <=> x>9/11 => threshold at 0.82
1>4 <=>  x<1/2   => threshold at 0.50 
2>3 <=>  x>17/42 => threshold at 0.40
2>4 <=>  x>1/7  => threshold at 0.14
3>4 <=>  x<2/3  => threshold at 0.67

So depending on your SVO, you rank these four options differently.


              4/2             1/2              3/2               1/4             3/4               3/1

|----------0.14---------0.26----------0.40----------0.50----------0.67----------0.82----------|


If 0.00{X{0.14, 4>2, 1>2, 3>2, 1>4, 3>4, 3>1 --->> 3>1>4>2, input should be [2413]
If 0.14{X{0.26, 4<2, 1>2, 3>2, 1>4, 3>4, 3>1 --->> 3>1>2>4, input should be [2314]
If 0.26{X{0.40, 3>2>1>4, input should be [3214]
If 0.40{X{0.50, 2>3>1>4, input should be [3124]
If 0.50{X{0.67, 2>3>4>1, input should be [4123]
If 0.67{X{0.80, 2>4>3>1, input should be [4132]
If 0.80{X, 2>4>1>3, input should be [3142]

Note that there are inputs by participants that are not consistent with choosing in accordance with SVO. For instance, the input [1234] suggests that the first option is preferred over the second (x<0.26) and the first over the third (x>0.82), which is a contradiction.

From the data analysis that we did at the time, I remember that our SVO-values were consistent with what was found using the other methods, including that 15 or 20% or so could not be properly classified.

For some reason, we thought that SVO-values might be larger than one as well, and also included:

1)  31 72
2)  22 80
3)  59 64
4)  41 67

1>2 <=> x<9/8 => 1.13
1>3 <=> x>7/2 => 3.50
1>4 <=>  x>2 => 2.00
2>3 <=>  x>37/16 => 2.31
2>4 <=>  x>19/13 => 1.46
3>4 <=>  x<6 => 6.00

but I don’t think we had many people with such high SVO’s.

A true comparison (within person?) would still be useful, I guess. But nevertheless, having this as an option might be good value to use in already long surveys. Running it twice, the second time with slightly different numbers, would also give nice test-retest values. So use at your own risk. If it is of use: credit where credit is due please!