A problem with soaring acceptance rates

Below follows Lingyun Lai’s first blogpost:

Since Mittins et al., in their book Attitudes to English usage, reported an overall acceptability of 50 English usage items in 1970, no systematic replication research had been conducted, until, from 2011 onwards, the Bridging the Unbridgeable project carried out a survey to repeat the original Mittins project. In the MA course Testing Prescriptivism Course, we were asked to compare the results of those two studies.

I imported the original polls’ results for the option “unacceptable under any circumstances” into Microsoft Excel 2013, and used the formula (100 – the unacceptable rate) to calculate the general acceptance rate. I similarlyimported the corresponding 1970 data and got the following results:

Lingyun Mittins

Figure 1. A comparison between 1970 and the Bridging the Unbrdgeable polls

The figures suggest that the acceptance rates for the ten usage problems studied increased drastically since forty years ago. The average rate had risen from 32.4% in 1970 to 95.9% forty years later. Does this trend imply a more tolerant attitude or a decline of prescriptivism in the new millennium?

Before jumping to a conclusion, I decided to explore the data further, and checked other statistics with SPSS 23. I noticed a big difference between the Standard Deviation values for two of the sentence polls, i.s. 12.1 versus 4.4. Standard Deviation is a statistical measure that allows us to observe the degree of variance among observations. My findings mean that the 1970s survey respondents varied their attitudes to different usage items significantly, whereas respondents of 2012 online survey demonstrated a much more homogenized attitude towards all items. See Figure 2 for this difference between the two groups of respondents:

Lingyun Mittins (2)Figure 2. Group Statistics of the 1970 and 2012 surveys.

Taking both types of results into consideration, I think there could be three possible interpretations:

  1. The results reflect a genuine picture. After forty years’ debate on prescriptivism, the general public has adopted a much more tolerant attitude towards almost all controversial language usages.
  2. The differences in variation are an indicator of two different populations (in a statistical sense). The 1970 and 2012 studies were conducted with two different methods: through onsite interviews and an online survey, so perhaps the Bridging the Unbridgeable polls only sampled a subgroup of the general public, the netizens, and perhaps this group of people accepted most of the controversial usages.
  3. The differences in variation are an indicator of the inefficiency of using online polls. A possible scenario could be that many online respondents did the survey rather casually, and responded all items with a similar pattern.

I would appreciate blog readers’ feedback on this issue!

Advertisements
This entry was posted in MA Leiden, polls and surveys and tagged . Bookmark the permalink.

2 Responses to A problem with soaring acceptance rates

  1. Carmen Ebner says:

    Interesting post! However, one should also not forget that Mittins et al.’s acceptability ratings are averaged ratings (p. 13). In order to compare the results, it is thus necessary to calculate the average of all acceptable contexts of the usage polls, which would result in lower acceptability ratings as the ones presented here. It is, however, very true that online polls and questionnaires are particularly difficult and that a self-selection bias does need to be taken in consideration. This bias does show when you look at the averaged ratings in comparison.

  2. adrianstenton says:

    Interesting indeed. Whilst I think that we do have to be very careful with the self-selection bias, I wonder if we might now also be “reading” the questions slightly differently. If I understand them correctly, the responders to Mittins et al.’s questions knew that they were being asked about points of grammar, and responded accordingly. One of the issues I have with Mittins et al. is that in some cases at least, e.g. Q24 “He only had one chapter to finish.”, there is no possibility of confusion if “only” is in the “wrong” place. If the example had been more like “John only saw the lion” (from Greenbaum & Whitcut, 1988, p. 496), the reader is more likely to wonder whether this means “he didn’t shoot the lion” or “he didn’t see the tiger”. As somebody who has responded to the BtU survey, and as a copy-editor, I would accept the Mittins et al. question, but query the Greenbaum & Whitcut sentence.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s