**3. Results and Discussion**

The pandemic changed the importance that some consumers gave to different food purchasing attributes. Our segmentation identifies three different groups of consumers attending to the importance of those changes. The first segment, with 38.6% of the sample, included consumers who had not modified the importance assigned to the food purchasing attributes before and during the lockdown. The second segment, with 47.5% of the sample was comprised by consumers who showed slightly modifications. Finally, the third segmen<sup>t</sup> of consumers, with 13.9% of the sample, included those who showed severe changes.

### *3.1. Amount Purchased of Di*ff*erent Food Items*

The variation of the amounts of different food items purchased by the consumer segments the week prior and after the state of alarm are shown on Table 3. This table shows the average scores for the amounts purchased before the state of alarm compared to those purchased in a normal situation. Thus, a one means that they buy much less than usual and a five means that they buy a lot more than usual. Therefore, scores over three indicate that more has been bought, and lower scores indicate that less has been bought.

In general, the food items purchased the most in the week prior to the state of alarm were rice, pasta and legumes, dairy products, meat, fresh fruit and/or vegetables, baked goods, and frozen foods. These are basic necessity foods and many of them are easily storable. Of these, in week 1 of the lockdown, the increase in dairy products, meat, fruit and vegetable and baked goods remained the same and the purchase of rice, pasta and legumes, and frozen products fell. The least purchased food items before the lockdown were soft drinks and juices, spices, condiments and sauces, snacks, bottled water, and beer, wine, and spirits. Of these, the purchase of beer and snacks increased in week 1. The purchase of bottled water and soft drinks remains the same. Lastly, the foods that remained the same before the lockdown are coffee and infusions, olive oil, fish, and canned products. All of the latter fell in week 1 of the lockdown. Regarding olive oil, an essential part of the Mediterranean diet, it was not added in a significant way when panic buying before the state of alarm, as all segments have ratings of around 3.


**Table 3.** Amount of different food groups purchased in the pre-state of alarm week and in which direction the change went in the first week post-state of alarm for each of the consumer segments.

1 Size of the segment; \*\*\*, \*\* and \* indicate significant differences with a 1%, 5%, and 10% margin of error; the asterisks shown on the food column reflect differences among segments. The asterisks shown in the segmen<sup>t</sup> column refer to the differences between the week prior and the first week after the state of alarm for each segment; the arrows mark the direction of the variation ↓ the amounts purchased decreased in week 2 compared to week 1. ↑ The amounts purchased increased in week 2 compared to week 1.

In segmen<sup>t</sup> 1, which comprised 38.6% of consumers, the most purchased foods in the pre-lockdown week were the same as for the total population, although the amount purchased was slightly lower in all cases: dairy products, baked goods, meat, fruit, rice, pasta, and legumes. Of these, in week 1 of the lockdown, only the purchase of meat increased, whereas rice, pasta and legumes decreased, with this fall being more significant than for other segments. Of the least bought products, snacks and beer, their purchase increased the following week, and it did so in greater proportion than for the total population. For the foods that remained the same, the amount purchased decreased in week 1 of the lockdown for oil, fish, and canned food, whereas for coffee it remained similar.

In segmen<sup>t</sup> 2, which comprised 47.5% of consumers, the food items that were purchased the most were similar to the total population, and the amounts purchased were slightly higher than the total population. The variation in week 1 of the lockdown was also similar. Regarding the products whose amount purchased dropped, they were also similar, and in week 1 the only group that increased were snacks, with the amount purchased for all other products remaining the same. For the food items that remain more stable, the only group that was purchased less was canned food, with all other categories remaining the same. This is the intermediate segment, which reacted by modifying the amounts purchased, but in a more moderate way than segments 1 and 3.

Lastly, segmen<sup>t</sup> 3, which comprised 13.9% of consumers, is the most sensitive to change. The increase in the amount purchased the week prior to the lockdown took place for the same groups of products as for the total population, but in this case, the increase was more pronounced. In week 1 of the lockdown, the amounts purchased were similar for all groups of products, except for rice, pasta, and legumes. The products purchased at a lesser rate were similar to the rest of the population, but it is worth noting that the purchase of beer, wine, and snacks was much lower during the week prior, and these levels were maintained in week 1 of the lockdown. This segmen<sup>t</sup> also purchased less spices, condiments, and sauces, and during week 1 they purchased even less of these. In the categories of more stable products, there was a decrease in the amounts of coffee, oil, and canned food purchased. The amount of fish purchased remained lower than normal but stable during that period of time.

### *3.2. Food Buying and Places of Purchase*

As regards shopping, a majority of consumers leave the home to do it, but this proportion was noticeably lower in segmen<sup>t</sup> 3, where the online or telephone option was much more prominent than for the other segments. When asked if they conducted these ways of shopping before COVID-19, the answers were very similar in all segments, and the percentage of those who did it often or on a regular basis was very low. However, this crisis boosted these ways of shopping, especially among consumers in segmen<sup>t</sup> 3. However, when asked whether they will continue using online shopping, in this segmen<sup>t</sup> there was a smaller percentage of consumers that will do so compared to others, where almost 30% of consumers said they will use the online channel to buy food. These results can be found in Table 4.


**Table 4.** Traditional shopping versus online food shopping for home.

1 Size of the segment; \*\*\* note significant differences with a 1% margin of error, respectively; Different letters in the same row mean significant differences for the segments (*p* < 0.05).

Regarding the places of purchase (Table 5), the scale used was an ordinal scale of five items, where 1 means that the person never purchased at that type of establishment, and a five means that they always did. The establishment of choice was the supermarket or hypermarket, followed by traditional shops, indoor markets and, lastly, buying straight from the producer, and there were hardly any differences among segments. Regarding the differences in behaviour between the week prior to the lockdown and during the lockdown, there were differences in almost all cases. The supermarket/hypermarket as the preferred choice increased mainly for segmen<sup>t</sup> 1, whereas it decreased for segmen<sup>t</sup> 3, although not significantly. All other places of purchase suffered a drop in importance.


**Table 5.** Place of purchase of food for home.

1 Size of the segment; \*\*\* and \* note significant differences with a 1% and 10% margin of error, respectively; the asterisks in the segmen<sup>t</sup> boxes refer to the change between considered periods (*t*-test of related samples). The asterisks on the variables column refer to the differences between segments (ANOVA).
