Actually, there are no strict rules for using a specific number of scales in the Likert scale. Whether you want many or few, even or odd, there's no prohibition, but there are indeed advantages and disadvantages.

Every time I'm asked whether even or odd is better, I can't simply answer, "it's up to you." I will share the characteristics of even and odd scales for this Likert scale.

As clearly seen, the difference between even and odd scales is the midpoint or center point. Odd scales definitely have a value in the middle, and the label usually attached is "neutral"—neither to the left nor to the right. There's something called a "safe answer" in odd scales. Even scales don't have that privilege. Because there's no value in the middle, even scales force respondents to choose left or right, even if it's only a one-point difference, the direction of the respondent's opinion is clear. There is no "safe answer" in even scales. Then, when to use even and when to use odd? I will share some theories I've gathered from various sources. Before going there, I want to share about the types of scale questions.

There are question types that ask about opinions or feedback, and there are also those that ask about desires or what respondents will do in the future. The first type asks about the past and present, the second type asks about possibilities in the future.

Examples of the first type include customer satisfaction, "How satisfied are you with our service or product?"—this is a question about experience and its relation to the past. This first type is suitable for even scales. Why? Because when respondents have already experienced something, they have to decide whether they are satisfied or not, like it or dislike it. There's no safe answer here. It might be a bit confusing to answer "neutral" when asked if you like the product.

An example of the second type is purchasing intention—the desire to buy, or the question, "How likely are you to buy or use this product if it launches?" Questions about future possibilities are more appropriately answered using an odd scale, because it's still hypothetical, and respondents might decide not to go left or right, i.e., neutral. For example, "I like the product, but I don't know the price yet, where it's sold, or how big the portion is." This means there are still other parameters to know before finally buying. If we look at the commonly used frameworks, I'll take CSI and NPS. CSI (Customer Satisfaction Index) uses an even scale [1-10], while NPS (Net Promoter Score), or how likely someone is to recommend, uses an odd scale [0-10]. Whether you use the same formula or not, it's fine as long as the data is accurate with the Likert scale.